Author’s Note: This essay draws on published research, the Nature of Fashion initiative’s work, and the Shadow Explorer framework I have been developing in this series — a set of questions designed to examine what any given system cannot see, not just what it can. That framework feels particularly necessary here, because the AI capabilities now entering the textile waste system are genuinely powerful, and the risks of overestimating them are genuinely significant. AI was used as a thinking partner.

What the Lagoon Is Doing

Off the coast of Ghana, in the Cor lagoon, something is happening that no engineer designed and no algorithm predicted.

Microbial communities living in water saturated with synthetic textile fibres — the runoff from one of the world’s largest second-hand clothing markets, fifteen million items arriving at Kantamanto market in Accra every week — are adapting. They are developing, through the slow pressure of evolutionary necessity, a biochemical capacity to process synthetic polymers that have existed in the environment for only decades. A site of acute environmental crisis is, in a very literal and biological sense, a site of evolution.

Katya Dodd, head of environmental justice for the OR Foundation, which works alongside these communities, describes what is happening in terms that resist the language of technological optimisation: “Innovation, pragmatism and justice are inseparable when we align with the intelligence that has sustained life for billions of years.”

By intelligence, she is pointing towards something broader than information processing or prediction. Ecological intelligence emerges through continual feedback between living organisms and the places they inhabit. It is knowledge generated through participation rather than abstraction, through adaptation rather than calculation. Every organism is constantly learning from the consequences of its actions in a specific environment, and over evolutionary time those lessons become embodied in forms, behaviours, and relationships.

The microbial communities in the Cor lagoon are expressing precisely this kind of intelligence. They are not solving an engineering problem. They are responding to environmental pressures in real time, participating in a living system whose conditions are changing around them.

This is the sentence I want to begin with. Not because it is the most technical observation in the room, but because it contains the thing that the most powerful AI systems currently entering the textile waste space are structurally unable to account for.

The Shadow Explorer Applied to a System

The Shadow Explorer framework asks, of any piece of information or any proposed solution: what are the assumptions it doesn’t state? Whose perspective isn’t in it? What context is missing? What does the source not know — or not want you to ask — about how this was produced?

Applied to artificial intelligence in the fashion system, this becomes a very specific set of questions. Not what can AI do — the capabilities are real and I will describe them fairly — but what kinds of reality does the system perceive, and what kinds of reality remain outside its frame?

This distinction matters because optimisation and understanding are not the same thing. A system can become extraordinarily good at improving measurable outcomes while remaining blind to the relationships that generate those outcomes in the first place. The question is not whether AI can process more information than humans. It plainly can. The question is whether processing information is equivalent to ecological understanding.

Asha Single, director of the Nature of Fashion initiative, frames the problem with unusual clarity. The fashion industry is missing the decomposer function: the biological layer that breaks complex matter back into molecular building blocks so that the cycle can begin again. Everything the industry produces ultimately escapes the system rather than re-entering it. The result is a design failure with consequences distributed catastrophically unevenly around the world — visible as landfill and toxic waterways in Ghana, invisible as a collection bin on a European high street.

Nature rarely solves this kind of problem through extraction alone. In healthy ecosystems, decomposers do far more than eliminate waste. They reconnect materials to the wider community of life. Fungi, bacteria, insects, and countless other organisms transform what appears to be an endpoint into the beginning of another cycle. Their work maintains resilience, preserves diversity, recovers nutrients, and creates conditions for future growth.

AI is now entering this system at several points. It can sort textile waste by fibre composition at 95% accuracy. It can predict the glucose yield of an enzymatic hydrolysis process from feedstock characterisation data. It can discover and design plastic-degrading enzymes — not merely finding ones that exist in nature, but generating novel ones computationally. It can optimise multi-step industrial cascades for maximum material value extraction. These capabilities are real, they are documented, and they are growing rapidly.

The Shadow Explorer question is not whether they work. It is: what does an AI system see when it looks at the fashion waste problem? And what does it not see?

What AI Sees

An AI system looking at the textile waste problem sees, with increasing precision, the following.

It sees fibre composition. Through hyperspectral imaging, Raman spectroscopy, and computer vision, it can characterise the molecular makeup of a mixed textile batch — the proportion of polyester, cotton, elastane, nylon — with greater speed and accuracy than any human sorter. SINTEF’s research, published this week, found that AI-enabled precise sorting could reduce CO2 emissions by up to 97% compared to incineration by directing material to appropriate processing streams.

It sees process yield. Given a feedstock characterisation, an AI system trained on industrial data can predict the glucose output of an enzymatic hydrolysis step, the PHA yield of a bacterial fermentation process, and the syngas composition from thermochemical gasification. It can recommend adjustments to temperature, enzyme blend, and processing time to improve output quality and reduce waste at each stage.

It sees enzyme candidates. AI protein language models can now classify plastic-degrading enzymes with 89% accuracy across diverse substrate types, and can generate entirely new enzyme sequences designed for specific degradation tasks. The search that Andrew Keys at Circle Economy conducted over three years — identifying local biotech partners with the right enzyme profile for their specific waste stream — is a process that AI can compress to weeks.

It sees material flows. At the system level, AI can model the movement of textile waste through regional collection, sorting, and processing infrastructure, identifying bottlenecks, optimising routing, and projecting the impact of different policy interventions on overall material recovery rates.

This is a formidable set of capabilities. Deployed well, it could meaningfully close the material loop in a fashion system that has operated without a decomposer layer for its entire industrial history. The case for AI in this space is genuine, and it deserves to be taken seriously on its own terms.

What AI Does Not See

And here is what the system cannot see.

It cannot see meaning. The microbial communities in the Cor lagoon are not optimising for efficiency in the way an industrial system does. They are responding to environmental pressures through countless interactions occurring across multiple scales of life. Over time, those interactions generate adaptive capacities that are inseparable from the conditions that produced them.

Nature’s innovations rarely emerge from the pursuit of a single objective. Living systems must simultaneously maintain resilience, adaptability, resource efficiency, reproductive success, and compatibility with the larger community of life. The biochemical abilities emerging in the lagoon are part of that broader context.

When Katya describes this process as intelligence, she is not using the word metaphorically. She is describing a form of adaptive knowing embedded in relationships. An AI system can identify patterns in enzyme databases, model degradation pathways, and generate candidate proteins. It can represent relationships as data. What it cannot do is participate in those relationships. It can model a lagoon. It cannot belong to one.

It cannot see justice. The consequences of the fashion system’s missing decomposer function are not evenly distributed. The eighty per cent of textile waste going to landfill and incineration does not land in the same places as the consumers who generated it.

Environmental injustice, like ecological collapse, emerges through relationships. Waste does not simply appear in one place and disappear in another. It moves through networks of trade, power, labour, policy, and consumption. The communities that bear the consequences are often far removed from the communities that receive the benefits.

Fifteen million items arrive at Kantamanto every week not because Accra chose to become the endpoint of the global fashion system, but because the system itself externalises costs geographically. An optimisation model can identify the most efficient pathway through a waste stream. It cannot independently ask whether the waste stream should exist in its current form, why its burdens fall where they do, or what a just distribution of responsibility might look like.

Just as an ecosystem cannot be understood apart from the relationships that sustain it, environmental justice cannot be understood apart from the relationships that distribute risk, benefit, responsibility, and power.

It cannot see relationships. Andrew Keys did not spend three years on the Circle Economy pilot because the enzyme platforms were hard to find. He spent three years because the pilot required something no database contains: businesses willing to operate non-competitively, to share risk with partners they didn't control, to remain committed to a vision whose commercial logic wasn't yet proven. What he was building, beneath the processing cascades, was trust. The industrial symbiosis that resulted is not a model that travels. It is an outcome that was grown — specific to those organisations, in that place, at that moment in how they understood what they were for. AI can compress the search for the right enzyme. It has no equivalent for the search for the right people.

It cannot see what it was not trained on. The populations generating the best quality, most longitudinal data for AI textile processing systems are the well-resourced recycling facilities in the Netherlands, Germany, China, and the United States. The waste streams in those facilities have specific characteristics: certain fibre blends, certain contamination profiles, certain collection and sorting histories. The AI models optimised for those streams will not perform equally well on the mixed, heavily contaminated, multiply-worn garments that arrive in Kantamanto — not because the technology is deficient, but because the training data does not reflect those conditions. We saw exactly this dynamic in the longevity medicine essay: algorithmic baselines calibrated on unrepresentative populations generate systematically misleading outputs for everyone else. The fashion waste problem is no different.

It cannot see what matters. An AI system optimising for material value recovery will maximise glucose yield, PHA production, and syngas quality. It will not ask whether the end products — the biodegradable bioplastic, the building block chemical — serve the communities living nearest the pollution, or whether they generate value for the biotech investors who funded the enzyme platform. It will not ask whether the definition of success encoded in its objective function — tonnes of waste diverted from incineration, CO2 equivalent saved — captures what the OR Foundation means by innovation, pragmatism and justice being inseparable.

The Two Pilots

The Nature of Fashion initiative is working across three pilots simultaneously, and the contrast between two of them illuminates this precisely.

The Circle Economy pilot will produce glucose, PHA biopolymers, and building block chemicals — materials with real market value, made possible in part by AI enzyme platforms developed by organisations that may retain the intellectual property behind them.

The OR Foundation pilot in Ghana carries a different logic entirely. It begins not with waste streams and process cascades but with ecological intelligence already present: the microbial communities adapting to their environment, the community members with generations of knowledge about the lagoon, the researchers and residents positioned as co-investigators rather than subjects of an intervention. Its ambition, as Asha describes it, is not to deploy a solution but to support and align with adaptive processes that are already happening — to ask what it means to work with intelligence that has sustained life for billions of years, rather than to replace it with intelligence designed last year.

No AI system can do what the second pilot is doing. Not because the technology is insufficiently advanced, but because what the second pilot is doing is not a task that can be represented as an optimisation problem. It is a practice of listening, of relationship, of co-investigation with communities whose knowledge the system has systematically excluded. The Shadow Explorer framework asks: whose perspective is missing from this information? The OR Foundation pilot is an attempt to answer that question by making those whose perspective is usually most absent the centre of the inquiry.

These two pilots need each other. The Circle Economy work demonstrates that it is technically and economically possible to build a decomposer layer in a specific regional industrial context. The OR Foundation work asks what it would mean for that layer to function in genuine alignment with the ecological and community intelligence that the technology alone cannot access. Neither is sufficient without the other.

The risk is that the first kind of work scales, and the second kind does not.

The Optimisation Trap

There is a pattern in how AI enters complex systems that is worth naming explicitly here, because it is not unique to fashion waste.

AI enters where the data is richest, and the objectives are most legible. It makes the measurable more efficient. It surfaces patterns in the quantifiable. It optimises for the criteria it is given. These are real and valuable contributions. They are also, structurally, a form of making the existing system faster rather than a form of questioning whether the existing system is the right one.

The fashion waste system has a measurable problem: too much material going to landfill and incineration. AI can help address that problem by improving sorting accuracy, designing better enzymes, and optimising processing cascades. If the objective is to divert tonnes of waste from incineration, AI is a powerful tool for reaching it.

But Asha’s question — what would the fashion industry look like if it functioned the way a natural ecosystem does? — is not asking how to optimise the existing system. It is asking whether the existing system has the right structure, the right relationships, the right understanding of what waste is and where it comes from and who bears its costs. It is a question about design, not efficiency. And the answer to a design question cannot be produced by an optimisation system, however powerful.

Natural ecosystems do not optimise. They diversify, adapt, develop relationships, build redundancy, maintain equilibrium through the constant co-evolution of their components. The decomposer function in a forest is not the fastest possible mechanism for breaking down leaf litter. It is the most embedded possible mechanism — the one that grew, over millions of years, in the specific conditions of that forest, in relationship with every other element of the system.

AI-designed decomposers can be fast and precise. Whether they can be embedded, in the sense that matters for genuine regenerativity, depends entirely on how they are developed, by whom, in relationship with whose knowledge, and accountable to whose definition of what the system is for.

The Questions Worth Asking

The Shadow Explorer protocol that I introduced in the health monitoring series asks, when confronted with a system that presents itself as authoritative: what triggered this response, and what context is missing? What does the system not know? Who benefits from my accepting the output rather than questioning it?

Applied to AI in the fashion waste system, the relevant questions are these.

Who is training the models? The most powerful AI sorting and enzyme discovery systems are being built by organisations with access to the richest data — the well-resourced facilities in high-income countries. The waste profiles, contamination levels, and processing conditions in those facilities are not representative of the global fashion waste problem. A model trained on Dutch recycling infrastructure may perform poorly on the conditions in Accra, or Lagos, or Dhaka, or any of the communities where the fashion system’s consequences are most severe.

What is the objective function? AI systems optimise for the criteria they are given. Tonnes of waste diverted, material value recovered, CO2 saved — these are legible, measurable, and genuinely important. They are not the same as: communities protected from pollution, workers in the sorting sector treated fairly, local industrial ecosystems developed rather than superseded by global platforms, the intelligence of the people who know their place best incorporated into the design of the solution.

Who owns the outputs? Andrew Keys is building a regional industrial symbiosis that produces glucose, PHA biopolymers, and building block chemicals. Those outputs have market value. The organisations that funded the AI enzyme platforms may own the intellectual property that makes the process possible. The distribution of that value — between the recycling facility, the biotech company, the regional government that funded early-stage work, and the communities whose waste created the feedstock — is not determined by the technology. It is determined by governance structures that are not yet in place.

What knowledge is being excluded? The OR Foundation pilot is asking what it means to build solutions in genuine partnership with the communities most affected. AI systems that enter the fashion waste space without this question are not neutral. They are replicating the same structural pattern that created the problem: decisions made in high-income countries, consequences borne in low-income ones, the knowledge of affected communities treated as irrelevant to the design of the response.

What Ecological Intelligence Actually Means

This is not an argument against AI. It is an argument against mistaking optimisation for wisdom. The challenge is not whether AI should be used. The challenge is whether its use remains accountable to forms of knowledge that originate outside the machine.

The deeper question raised by the Cor lagoon is not whether artificial intelligence can help us build a circular economy. It almost certainly can.

The question is whether we are willing to recognise that the most sophisticated forms of intelligence involved in that transition may not be artificial at all.

In the waters of the lagoon, microbial communities are responding to a crisis not through optimisation but through participation. They are learning from the conditions around them, adapting to them, and in doing so becoming part of an ongoing evolutionary conversation that began long before humans arrived and will continue long after our current technologies are obsolete.

AI can help us analyse that conversation. It can help us model it, accelerate aspects of it, and perhaps even support it.

But it cannot replace it.

The microbes in the lagoon are not asking how to optimise a waste stream. They are responding to the realities of a place. The question for us is whether our technologies will learn to support that living conversation — or whether, in our fascination with optimisation, we will lose sight of the relationships that make adaptation, resilience, and regeneration possible in the first place.

References and Further Reading

Nature of Fashion initiative — Biomimicry Institute https://biomimicry.org/nature-of-fashion/

OR Foundation — Ghana

https://theor.org/

Circle Economy — Netherlands

https://www.circle-economy.com/

AI sorting — SINTEF, 97% CO2 reduction potential “AI Sorting Could Turn Discarded Clothes into New Fabrics More Efficiently.” TechXplore, June 4, 2026. https://techxplore.com/news/2026-06-ai-discarded-fabrics-efficiently.html

AI enzyme discovery — protein language models Medina-Ortiz D, et al. “Protein Language Models Accelerate the Discovery of Plastic-Degrading Enzymes.” bioRxiv, February 2025. https://www.biorxiv.org/content/10.1101/2025.02.09.637306v1

Shadow Explorer framework The Architecture Beneath Your Thinking · Teaching Children to Think in Shadows

Kantamanto market — 15 million items per week OR Foundation / Circle Economy reporting on Accra textile waste.

https://theor.org/

Disclaimer: This essay is analytical and exploratory. It does not constitute professional environmental or technical advice.

Humans are a species shaped in relationship with landscapes. For most of our history, the mind developed not in front of screens, but in the presence of wind through leaves, shifting light, flowing water, and the quiet movements of other living beings. These environments were not silent; they were rich with information, but they did not overwhelm.

Like all organisms, we evolved within a specific set of conditions. Our sensory systems, our patterns of attention, even our rhythms of fatigue and recovery, emerged in conversation with the natural world. In those environments, attention was not a constant strain. It moved fluidly, sometimes focused, sometimes open, always in dialogue with complex surroundings that were not chaotic.

Today, many of us live in environments that ask something different of the mind. We hold focus for long periods, filter competing signals, and manage a steady stream of abstract information. This shift places continuous demand on a particular kind of attention—one that, like any biological system, is not designed to operate without rest.

The Strategy

The human brain uses multiple forms of attention, but one of the most energy-intensive is directed attention, the ability to focus deliberately while suppressing distractions. This is the attention used for reading, problem-solving, and navigating modern digital environments. It is powerful, but costly. Over time, it fatigues.

In the 1980s, researchers proposed Attention Restoration Theory (ART), suggesting that certain environments allow this fatigued system to recover. The key lies in a different mode of attention: what they called soft fascination.

Soft fascination occurs when attention is gently captured without effort. A moving stream, the shimmer of leaves, the rhythm of waves, these hold the mind, but do not demand control. Unlike directed attention, they do not require the suppression of competing stimuli. Instead, they allow the brain’s effortful systems to rest while attention remains lightly engaged.

Natural environments consistently provide this kind of stimulation. They tend to share four qualities that support recovery:

• Soft fascination: patterns that attract attention without overwhelming it

• Being away: a sense of psychological distance from demands

• Coherence: structured, legible environments that the brain can easily organise

• Compatibility: freedom to move and engage at one’s own pace

In contrast, many urban environments present fragmented, competing signals—noise, motion, alerts—that require continuous filtering. Rather than allowing attention to rest, they prolong their effort.

From a biological perspective, this difference matters. Living systems operate through cycles, effort followed by recovery. Muscles fatigue and rebuild. Sleep restores neural function. Ecosystems regenerate after disturbance. Attention appears to follow a similar pattern.

Natural environments may support this cycle not only because they are less demanding, but because they are patterned in ways the brain processes efficiently. Many natural forms, tree branches, river networks, leaf veins, share fractal structures, repeating patterns across scales. Research suggests the visual system can process these patterns with less effort, possibly because similar structures appear throughout the body, including in neural networks themselves.

In this way, nature offers a rare combination: high richness of information with low cognitive cost.

The Potential

If natural environments help restore attention, they are not a luxury; instead, they are part of the conditions under which clear thinking becomes possible.

This insight invites a shift in how we design the spaces we inhabit. Instead of treating nature as decoration, we can see it as infrastructure for cognitive function.

• Workplaces can integrate views of vegetation, natural light, and quiet outdoor access to support sustained focus

• Cities can prioritise accessible green spaces, not only for recreation but for mental recovery

• Schools can incorporate outdoor learning and nature exposure to support attention and learning capacity

• Healthcare settings can use natural elements to aid cognitive and emotional restoration

Even at a small scale, the pattern holds. A short walk among trees, a view of water, or the presence of plants can begin to interrupt cycles of mental fatigue.

For individuals, the lesson is simple but profound: attention is not only managed internally. It is shaped by the environments we move through. By choosing conditions that allow the mind to recover, we align with a deeper principle found across living systems—that resilience depends not on constant output, but on the rhythm between effort and renewal.

In reconnecting with these environments, we are not stepping away from productivity. We are returning to the conditions that make it possible.

Sources:

• Kaplan, R. et al. “Attention Restoration Theory: Exploring the Role of Soft Fascination.” Environment and Behavior
  https://www.ovid.com/journals/envab/fulltext/10.1177/0013916518774400~attention-restoration-theory-exploring-the-role-of-soft

• Berman et al. “Attention in Urban and Natural Environments.” Current Directions in Psychological Science (via PMC)
  https://pmc.ncbi.nlm.nih.gov/articles/PMC6430180/

• Zhang et al. “Restorative effects of small urban parks.” Frontiers in Public Health
  https://pmc.ncbi.nlm.nih.gov/articles/PMC12683676/

• Basu et al. “Attention Restoration Theory.” Environment and Behavior
  https://doi.org/10.1177/0013916518774400

• Neilson et al. “A review of the limitations of Attention Restoration Theory.” Visions for Sustainability
  https://doi.org/10.13135/2384-8677/3323

• Ohly et al. “Nature’s broken path to restoration.” Journal of Environmental Psychology
  https://doi.org/10.1016/j.jenvp.2018.08.006

• Frontiers in Forests and Global Change (2025) – Restorative environments and neural mechanisms
  https://www.frontiersin.org/articles/10.3389/ffgc.2025.1651800

AskNature Chat was used as a Thinking Partner.

This might seem like an unlikely place to look for insights about mental resilience. But the science of the gut–brain axis — the communication network linking the digestive system, immune system, nervous system, and endocrine system — is beginning to suggest that the ecosystems living inside us shape not only our physical health but also how we think, feel, and adapt.

The body as an ecosystem

Trillions of microorganisms — bacteria, fungi, and other microscopic life — inhabit the gut, skin, and many other parts of the body. Collectively, this community is known as the microbiome.

One of the central ideas in systems thinking is that living systems rarely operate in isolation. Forests, coral reefs, and wetlands depend on networks of cooperation, competition, and nutrient exchange.

The human body may work similarly. Rather than being a single organism acting alone, humans can be thought of as host ecosystems — environments in which many species coexist.

The gut microbiome is one of the most densely populated of these ecosystems. It contains hundreds of microbial species that perform functions supporting digestion, immunity, and metabolic regulation.

When this ecosystem is balanced and diverse, it tends to function smoothly. When it becomes disrupted — through illness, stress, poor diet, or environmental factors — problems can emerge. Researchers sometimes refer to this imbalance as dysbiosis, and increasingly it appears linked not only to physical health conditions but also to aspects of mental well-being.

The gut–brain conversation

Communication between the gut and the brain happens through several overlapping pathways.

One of the most important is the vagus nerve, a major nerve that connects the brain to organs throughout the body. But neural signals are only part of the story.

Microorganisms in the gut also produce a wide range of chemical compounds that influence physiology. Some microbes help synthesise neurotransmitters or precursors involved in mood regulation, including substances related to serotonin, dopamine, and gamma-aminobutyric acid (GABA). Others produce short-chain fatty acids, molecules created during the fermentation of dietary fibre, which can influence inflammation, immune activity, and aspects of brain function.

The immune system also acts as a communication channel. Because a large proportion of the body’s immune cells reside in the gut, microbial activity can influence inflammatory responses throughout the body.

Taken together, these pathways create a complex, two-way dialogue between microbes and the brain.

Microbiome diversity and resilience

One of the most interesting patterns observed in many ecosystems is the relationship between diversity and resilience. In forests or coral reefs, diverse communities of organisms often make ecosystems more stable — if one species declines, others may compensate.

Researchers suspect something similar may occur in the microbiome. A diverse microbial community appears to be associated with healthier metabolic regulation and stronger resistance to disturbances. Some studies have also suggested links between microbiome diversity and psychological well-being, though this area of research is still developing.

Stress, diet, antibiotic use, sleep disruption, and environmental exposures can all influence microbial diversity. Because these same factors also influence mental health, teasing apart cause and effect remains a challenge. Still, the emerging picture suggests that supporting microbial diversity may be one component of supporting broader physiological resilience.

Stress, inflammation, and the microbiome

Mental resilience is often discussed in psychological terms — the ability to cope with challenges, recover from setbacks, and adapt to change. But resilience also has biological dimensions.

Chronic stress can influence the microbiome in several ways. Stress hormones can alter gut motility, immune responses, and the environment in which microbes live. Over time, these changes may shift microbial communities. And disruptions in the microbiome may in turn influence levels of inflammation in the body.

Inflammation has been linked to a wide range of conditions affecting both physical and mental health. Rather than a simple cause-and-effect relationship, the gut–brain axis likely represents a feedback system: stress affects the microbiome, the microbiome influences physiological responses to stress, and the cycle continues.

What shapes the microbiome?

Because the microbiome behaves like an ecosystem, it is influenced by environmental conditions.

Diet is one of the strongest. Diets rich in plant fibres provide fuel for microbes that produce beneficial short-chain fatty acids, while highly processed diets low in fibre may reduce microbial diversity. Antibiotics can dramatically alter microbial populations — often necessary for treating infections, but disruptive to microbial ecosystems when used repeatedly or in broad-spectrum form.

Early childhood also appears to be a formative period, with birth method, early diet, and environmental exposures all shaping microbial communities in lasting ways.

Perhaps the most intriguing influence is contact with nature. Soil, plants, and outdoor environments contain diverse microbial communities, and exposure to these environments may introduce beneficial microorganisms and help support diversity. This idea — sometimes called the biodiversity hypothesis — suggests that reduced contact with natural ecosystems may influence immune regulation and health in ways we are only beginning to understand.

A design question for modern environments

Urban lifestyles often combine highly processed foods, reduced contact with soil and natural environments, frequent antibiotic exposure, and high levels of psychological stress. Each of these factors may influence microbial ecosystems in ways that compound over time.

This raises broader questions from a design perspective. How might we create environments that support healthy microbial diversity? Could access to green spaces, varied diets, and contact with natural ecosystems help maintain more resilient internal microbiomes?

These questions connect biology, urban design, and environmental health in unexpected ways.

Supporting the internal ecosystem

The science of the microbiome is still evolving, and many questions remain unanswered. However, several general patterns are already reasonably well supported.

Practices that tend to support microbial diversity include eating a wide variety of plant-based foods, consuming dietary fibre from whole foods, spending time in natural environments, maintaining regular sleep, and managing chronic stress. These behaviours overlap strongly with recommendations for cardiovascular, metabolic, and cognitive health — a convergence that is a recurring theme in biology. Systems that support one part of the body often support many others.

The brain does not operate alone. It is connected to immune responses, metabolic signals, microbial ecosystems, and environmental inputs. Resilience may emerge partly from the stability of these interconnected systems — and caring for the ecosystems within us may be one more way of caring for the mind.

This article is part of The Connected Body, a series exploring what living systems can teach us about cognition, health, and the environments we design. Future essays will explore why walking improves thinking, how sleep restores cognitive function, and how natural environments support attention.

Subscribe now

The microbial garden beneath our feet.

AI-generated image.

AI was used as a thinking partner.

Cardiologists focused on the heart. Neurologists studied the brain. Endocrinologists looked at hormones. Each system had its own specialists, measurements, and treatments.

But biology has never really worked that way. Living systems are integrated. Circulation influences cognition. Sleep affects metabolism. The immune system shapes brain function. When one part of the system weakens, the effects ripple outward.

One of the most interesting examples of this integration is the relationship between the heart and the brain.

At first glance, they seem like very different organs. One pumps blood; the other generates thought. But in reality, they depend deeply on one another.

Your brain accounts for only about 2% of your body weight, yet it uses roughly 20% of the blood your heart pumps.[1] Every thought, memory, and decision depends on a vast network of blood vessels delivering oxygen and nutrients to billions of neurones.

Protect those vessels, and you support the brain. Damage them, and cognition itself can suffer.

This idea — that the health of the brain depends on the health of the body’s infrastructure — is becoming increasingly clear in modern research.

But it is also something nature has always known.

A systems lesson from biology

One of the first lessons in biomimicry is that nature rarely designs isolated solutions. Instead, biological systems evolve to support whole-organism resilience. Circulation, metabolism, immunity, and cognition are not independent modules. They are deeply interconnected processes that help an organism survive in its environment.

In many ways, the human body behaves less like a machine and more like an ecosystem. Energy flows. Signals travel. Resources move through networks. The stability of the system depends on the health of its connections.

When researchers study the link between cardiovascular health and cognitive decline, they are essentially rediscovering this systems principle.

Damage to blood vessels — through inflammation, pollution exposure, poor sleep, or metabolic stress — can disrupt the delicate network that nourishes the brain. And when that network degrades, cognition can degrade as well.

The infrastructure of thinking

We often imagine thinking as something that happens purely inside the brain. But biological intelligence depends on infrastructure.

Neurones require a constant supply of oxygen and glucose. Waste products must be cleared away. Temperature, sleep cycles, and immune signals all influence how well the brain functions.

If this support system falters, the brain struggles to operate effectively.

This insight mirrors something researchers increasingly recognise in artificial intelligence.

Large AI systems are often described in terms of algorithms and models. But their capabilities depend heavily on physical infrastructure — energy supply, cooling systems, specialised chips, and data networks. Without those supporting systems, the intelligence simply cannot run.

Biological cognition works similarly. The brain does not think in isolation. It thinks as part of a living system.

The two biological processes shaping both heart and brain

Many different stressors affect human health, but they often converge through two common pathways.

Chronic inflammation

Inflammation is the body’s natural repair response. In the short term, it protects us from injury and infection.

But when inflammation becomes persistent — triggered by stress, pollution, poor diet, or disrupted sleep — it begins to damage tissues.[2] Over time, this low-grade inflammation can injure the lining of blood vessels, accelerate plaque formation in arteries, and affect the brain’s protective barriers.

Markers of inflammation associated with cardiovascular disease are also frequently elevated in people experiencing cognitive decline.[3] The same process is affecting multiple systems at once.

Oxidative stress

Another shared pathway is oxidative stress.

Every cell produces unstable molecules called free radicals as part of normal metabolism. Under healthy conditions, the body neutralises them with antioxidants.

But environmental pollutants, ultraviolet radiation, psychological stress, and metabolic strain can tip the balance.

When this happens, oxidative stress damages cells, blood vessels, and mitochondria — the energy systems inside cells.

Both cardiovascular disease and neurodegenerative disorders are strongly associated with this kind of cellular damage.

Again, the mechanisms overlap.

The environments shaping our biology

If the body behaves like an ecosystem, then it is not surprising that the external environment also shapes our health.

Recent cardiovascular research increasingly recognises environmental stressors as significant contributors to disease risk.[4] These include air pollution, chronic noise exposure, extreme heat, chemical pollutants, and long-term psychological stress. Air pollution alone ranked fourth among global risk factors for mortality in the 2019 Global Burden of Disease analysis.[5]

All of these pressures influence the same biological pathways: inflammation, oxidative stress, and vascular damage.

Because the brain relies so heavily on the body’s vascular network, these environmental forces can also influence cognitive ageing.

A design problem hiding in plain sight

Seen from this perspective, health is not just a medical issue. It is also a design challenge.

How we design our environments affects the biological systems we depend on.

Urban air quality affects vascular inflammation. Noise affects sleep. Green space affects stress and recovery.

These factors influence the infrastructure that supports cognition itself.

Biomimicry encourages designers to look at how nature creates resilient systems. One recurring strategy is the creation of conditions that allow healthy processes to emerge naturally.

Ecosystems do not micromanage individual organisms. Instead, they create environments where life can flourish.

The same principle may apply to human health.

The everyday behaviours that support the system

Fortunately, the behaviours that support cardiovascular health also tend to support the brain.

Research repeatedly highlights a small set of habits that influence both systems.

Regular movement

Physical activity improves circulation, reduces inflammation, and increases blood flow to the brain. It also promotes neuroplasticity and hippocampal growth — the brain region most critical for memory and learning.[6]

Consistent sleep

During sleep, the brain clears metabolic waste and restores neural function. Research has shown that sleep is associated with a striking increase in the clearance of beta-amyloid, a protein implicated in Alzheimer’s disease.[7]

Whole-food dietary patterns

Diets rich in vegetables, whole grains, fish, nuts, and healthy fats support vascular health and reduce inflammatory stress.

Stress regulation

Practices that calm the nervous system — exercise, time in nature, reflection, or mindfulness — help prevent chronic activation of the stress response.

None of these interventions is dramatic. But over the years, small behaviours shape the conditions in which our biological systems operate.

Thinking like nature

One reason biomimicry is so powerful is that it changes how we frame problems.

Instead of asking how we fix a single issue, it asks: what conditions allow the whole system to work well?

Applied to health, this perspective shifts the focus from isolated symptoms to the environments and behaviours that support the body’s natural resilience.

The emerging science connecting heart health and brain health is a reminder that cognition itself depends on the health of these underlying systems.

Our bodies evolved to function within particular environmental conditions — movement, daylight, varied nutrition, and connection to the natural world.

When those conditions are present, many biological processes regulate themselves remarkably well.

When they are absent, problems begin to accumulate.

The bigger lesson

If there is one insight from living systems worth remembering, it is this: intelligence is supported by networks.

• In the brain, those networks are neurones.

• In the body, they are blood vessels and metabolic pathways.

• In ecosystems, they are flows of energy and resources.

Damage the network, and the system struggles. Support the network, and complex behaviour becomes possible.

Protecting cardiovascular health, it turns out, may also be one of the most important ways to protect the brain.

And the environments we create around us — our cities, workplaces, and daily habits — play a larger role in that process than we often realise.

A new series: Lessons from the Connected Body

This article is the first in a series exploring how insights from biology can inform how we think about cognition, design, and human environments.

Future pieces will explore why walking often improves thinking, how sleep supports cognitive function, why time in nature affects attention and mood, what the microbiome might teach us about mental resilience, and how environmental design influences learning and creativity.

Biology is full of systems that have evolved to solve complex problems.

Looking closely at those systems may offer clues for designing healthier environments — and perhaps even more thoughtful technologies.

The treasure hunt continues.

References

[1]The Cerebral Circulation, Cipolla MJ. San Rafael (CA): Morgan & Claypool Life Sciences, 2009. Available at: https://www.ncbi.nlm.nih.gov/books/NBK53094/

[2]Ekblom-Bak E et al. ‘Cardiorespiratory fitness and inflammation in Swedish adults: a cross-sectional study’, Neurology, 2022. See also: Kivipelto M et al. ‘Midlife vascular risk factors and Alzheimer’s disease in later life’, BMJ, 2001.

[3]Ridker PM et al. ‘C-reactive protein, the metabolic syndrome, and risk of incident cardiovascular events’, Circulation, 2003. For the brain–inflammation link: Ageing Research Reviews, 2024 review on inflammation, oxidative stress and cognitive decline.

[4]ESC / ACC / AHA / WHF Joint Statement on Environmental Stressors and Cardiovascular Disease, Circulation, 2025. Available at: https://www.ahajournals.org/doi/10.1161/CIRCULATIONAHA.125.079034

[5]GBD 2019 Risk Factors Collaborators. ‘Global burden of 87 risk factors in 204 countries and territories, 1990–2019’, The Lancet, 2020; 396(10258): 1223–1249.

[6]Jimenez-Maldonado A et al. ‘Neuroprotective mechanisms of exercise: a review’, The Lancet, 2025. See also: Erickson KI et al. ‘Exercise training increases size of hippocampus and improves memory’, PNAS, 2011; 108(7): 3017–3022.

[7]Xie L et al. ‘Sleep Drives Metabolite Clearance from the Adult Brain’, Science, 2013; 342(6156): 373–377. Available at: https://www.science.org/doi/10.1126/science.1241224

AI was used as a Thinking Partner

In November 2023, Google DeepMind introduced GraphCast, an AI model capable of producing a ten-day global weather forecast in under one minute, outperforming traditional physics-based systems across most evaluation targets. A year later, Microsoft’s Aurora surpassed both conventional and AI systems on more than 90 per cent of forecasting metrics.

These are not incremental improvements. They signal a structural shift in how we model planetary systems.

For policymakers, investors, and technologists, the implication seems obvious: if AI can forecast weather with unprecedented speed and accuracy, surely it can accelerate our response to climate change.

But beneath that optimism lies a more foundational question — one that has less to do with processing power than with what kind of intelligence climate reality actually requires.

What Climate Science Actually Requires

Climate science is slow by necessity.

The IPCC process takes years, not because scientists are inefficient, but because the system being studied is planetary-scale, nonlinear, and filled with feedback loops that resist simplification. Earth system models do not produce a single answer. They generate ensembles — families of plausible futures under varying assumptions and emissions trajectories.

The goal is not certainty. It is mapping the possibility space.

This posture mirrors how resilient ecosystems function. A forest does not optimise for one forecasted future. It maintains species diversity, functional redundancy, and modular organisation. Some trees thrive in drought; others in wet years. When disturbance comes — fire, pest, storm — the system absorbs the shock not because it precisely predicted the event, but because it preserved its adaptive capacity.

The Architecture of False Confidence

Most AI systems, however, are not built on ecological logic.

They are trained to collapse divergence into coherent synthesis. Helpfulness is rewarded. Hesitation is penalised. Fluency becomes a proxy for reliability. The result is an interface that feels decisive — even when the system being modelled is not.

When a policymaker asks, “What is the best carbon removal strategy for Southeast Asia by 2040?”, the framing assumes there is a best answer. The model responds within that frame. It narrows the possibility space. It synthesises.

That narrowing is not malicious. It is architectural.

In ecology, monocultures are efficient but brittle. A field planted with a single crop yields abundantly under stable conditions — but it collapses when stress arrives. Diversity is not inefficiency. It is insurance.

If AI becomes the primary interface between complex climate models and human decision-makers, it risks functioning as a cognitive monoculture — streamlining diverse model outputs into a single narrative, converting the IPCC’s carefully calibrated ensemble of futures into the false reassurance of a single projected path.

The danger is not inaccuracy. It is over-coherence.

Earth’s systems contain tipping points, nonlinear thresholds, and cascading feedback loops. An intelligence optimised for confidence may inadvertently erode the very habits of mind required to navigate such terrain.

Pattern Recognition Is Not Process Understanding

This distinction becomes critical when we examine what AI weather systems are actually doing.

GraphCast and Aurora are trained on vast reanalysis datasets — statistical reconstructions of atmospheric history. They detect structure within observed dynamics and extrapolate forward. For short-term forecasting, this works remarkably well. The results speak for themselves.

But climate change is not merely an extension of historical variability. It introduces boundary conditions without precedent: novel emissions trajectories, altered ocean circulation patterns, and destabilising ice-sheet feedbacks. These are not events that past atmospheric data has seen. When the system reorganises — and climate systems do reorganise — a model that has learned to be accurate within its training distribution may confidently extrapolate into territory where its confidence is no longer earned.

Pattern recognition excels within stable regimes. It is fragile at thresholds.

This is why Earth system models encode physical processes rather than just statistical regularities. The physics doesn’t change when the climate does. The historical pattern does.

Living systems balance pattern sensitivity with mechanistic constraint. Evolution experiments, but always within boundaries imposed by thermodynamics and ecological processes. Organisms survive not because they are optimised for yesterday’s conditions, but because they retain the capacity to respond when conditions shift.

From Forecasting to Regeneration

For decades, the central questions of climate science have been predictive: How much warming? How much sea-level rise? How much carbon must be removed?

Forecasting is essential. But forecasting alone does not redesign systems.

Consider what happened in the Loess Plateau of China — one of the most eroded landscapes on Earth, stripped of vegetation over centuries of overgrazing and farming. Beginning in the 1990s, a restoration programme introduced contour farming, native replanting, and grazing controls based on real-time monitoring of soil and vegetation recovery. The plateau regenerated. Rainfall retention improved. Biodiversity returned. Agricultural productivity rose. The intervention worked not because it predicted a single future but because it built adaptive feedback loops into the design — adjusting strategies as the landscape responded.

This is what regenerative intelligence looks like in practice. Not a dashboard that outputs the optimal solution. A system that monitors, responds, and recalibrates continuously as the living system evolves.

This is where AI’s second contribution to climate work must lie. If its first revolution is computational speed — doing what GraphCast does, producing forecasts faster and at lower cost than physics-based models — its second must be regenerative design: helping us build landscapes, cities, supply chains, and energy systems that behave more like ecosystems.

Design Principles for Ecological AI

If AI is to enhance climate intelligence rather than compress it, its architecture must mirror ecological principles.

Preserve diversity before optimising. Generate multiple viable pathways before ranking them. Surface trade-offs. Avoid single-answer outputs in complex systems contexts. Uncertainty is not a presentation problem to be smoothed away — it is the most important signal in the data.

Make uncertainty visible. Display confidence prominently. Identify sparse data zones. Flag embedded assumptions. The IPCC uses carefully calibrated language — likely, very likely, virtually certain — for good reason. AI interfaces operating in this domain should do the same.

Build for continuous feedback. Integrate ecological monitoring — satellites, soil sensors, biodiversity metrics — into dynamic recalibration loops. Static optimisation is brittle. Adaptive feedback is resilient. The Loess Plateau succeeded because the intervention was designed to learn from the landscape’s response, not to execute a fixed plan.

Prioritise long-term stability. Evaluate policies not only by immediate emissions reductions but by biodiversity integrity, ecosystem health, and resilience under stress scenarios. Optimising for a single metric in a complex system has a consistent historical track record of producing unintended consequences.

Distribute intelligence. Empower regional and community-level modelling tools. Resilience increases when intelligence is distributed rather than centralised. A single powerful model serving all geographies is, again, a monoculture.

Design for regeneration. Use AI not only to mitigate harm but to optimise soil recovery, habitat restoration, circular material flows, and adaptive urban design. Ask not only how much damage we can prevent, but how much adaptive capacity we can build.

These are not technical upgrades. They are philosophical shifts — in what we ask AI to do, and in what we reward it for doing well.

Intelligence Designed for a Living Planet

We are not simply building tools. We are constructing cognitive infrastructure — the systems through which future generations will interpret planetary risk and make decisions that will shape the climate for centuries.

Earth’s intelligence is distributed. It is iterative. It tolerates uncertainty. It branches possibilities before converging on survival pathways.

Climate science has slowly, painstakingly evolved to mirror that posture — to hold complexity rather than collapse it, to maintain ensembles rather than select winners.

Our AI systems must do the same.

Designed as monocultures — optimised for coherence, speed, and decisiveness — they will generate clarity at the expense of resilience. Designed ecologically — surfacing divergence before synthesis, embedding feedback, preserving adaptive capacity — they can amplify our ability to participate intelligently in planetary systems.

GraphCast and Aurora are genuinely remarkable achievements. The question is not whether to use them. It is what we build next, and what values we encode into that architecture.

Climate change is not simply an engineering problem. It is an adaptive systems challenge.

Adaptive systems do not survive by being certain. They survive by being able to adjust.

The intelligence we build should work the same way.

AI was used as a ThinkPartner

A child looks up from their work, uncertain. Something doesn’t fit. The plan assumed sunshine, but it is raining. The blueprint forgot about the neighbours. The answer arrived too quickly, and something feels off.

That moment of noticing — of sensing the edge of what you understood — is where real thinking begins.

Modern environments are very good at eliminating it.

What Optimised Spaces Actually Teach

Most kindergartens are designed for control. Constant temperature. Predictable light. Surfaces that clean easily, not surfaces that teach anything. The implicit curriculum is:

In my post on teaching thinking in the age of AI, I wrote about the danger of what I called “borrowed competence” — children who produce polished work by delegating the thinking to AI, and who gradually lose the internal capacity to think independently. The solution I proposed was not better techniques, but better noticing: helping children develop metacognitive awareness of what they don’t know, what the answer assumes, and what is outside the frame.

What I didn’t say then is this: noticing is not primarily a cognitive skill. It is first a sensory habit.

And sensory habits are formed — or extinguished — by the spaces we inhabit.

The Sketch

In 2024, I decided to design a kindergarten - just for fun!

Not a finished architectural proposal — a sketch. A thinking-out-loud on paper about what it might mean to design a space that teaches children to pay attention to living things.

The indoor plan has a central teaching area built around overlapping circular forms, like cells in a living organism, like water moving around stones. There is a Library Tower. A Kitchen. A Creative Room. In-between areas that don’t quite belong to any single zone — threshold spaces where children transition and, perhaps, notice the transition.

The outdoor plan has more to say. A Pollinator Garden. A Wildlife Garden. Solar Botanic Trees that rotate slowly with the sun. A Mud Kitchen. A Water Play area. A Sand Pit. An Upcycling Corner. A Beehive Climbing Structure that earns its name.

None of these spaces explains itself.

The pollinator garden is empty in February and buzzing in July. The solar trees track a rhythm that children don’t control. The mud kitchen produces results that depend on yesterday’s rain. The wildlife garden changes not because a teacher planned a lesson, but because something moved in and something else moved on.

This is not decoration. This is designed with an epistemological intention.

When the Space Asks the Questions

In my Shadow Explorer work with children and AI, the key intervention is simple: after a child receives an answer from a textbook, a teacher, or an AI, a single question appears.

What would this need in real life?

That’s it. No technique to master. Just a gentle interruption pointing to the edge of the idea.

What I am beginning to understand is that a well-designed biophilic environment asks this question constantly, without any adult present.

A child plants a seed and waters it once. Two weeks later, it is limp. What did this need that I didn’t give it?

A child builds a den in the willow structure and comes back the next morning to find it partly dismantled by the wind. What did I assume about the weather?

A child watches a bee ignore the flowerbed that the teacher called a “pollinator garden” because the blooms aren’t open yet. What did the plan forget about timing?

The environment is not explaining things. It is not narrating. It is not sending a notification. It is simply behaving like a living system — with consequences, dependencies, and the occasional surprise — and inviting children to make sense of it.

This is what I call “restorative presence.” But presence requires something to be present to. And that something cannot be a managed, optimised, artificially stable environment.

It has to be alive, which means: sometimes inconvenient, sometimes confusing, and sometimes in need of help.

Biomimicry as Pedagogy

The materials I am drawn to for the concept of a kindergarten were not chosen primarily for sustainability, though they are sustainable. They were chosen because they demonstrate biological principles that children can observe and question.

Mycelium insulation that grows into its mould, then stops. Lotus-effect renders on the outdoor walls that bead water and dry themselves. Structural forms that follow the branching logic of trees. Solar spheres that catch light from multiple angles, like eyes.

These are not novelties. They are legible lessons — but only if children are given the time and the silence to notice them.

My warning in the kindergarten design piece is exactly right: if every leaf reports data to a smartphone, we teach children that nature only matters when translated into numbers. The goal is not an instrumented nature. The goal is encountered in nature.

The child who watches condensation trace a pattern on a mycelium wall, and wonders why that corner is always wetter, is not following a curriculum. They are doing something more important: they are noticing an edge in their own understanding. They are discovering that they don’t know something they assumed they did.

That is metacognition. And it happened because the wall behaved unexpectedly.

Graduated Exposure, Not Protection

There is a version of biophilic kindergarten design that fails the children it claims to serve. It is the version that brings nature inside and tames it — plastic plants, projected forest sounds, a “nature corner” with pine cones in a basket.

This is nature as symbol. It teaches children that nature is something you look at, not something you live with.

The ecosystem post I drew on earlier makes the point precisely: nature does not eliminate challenge. It scaffolds it. The forest canopy moderates temperature so seedlings can establish before facing full intensity. It does not eliminate variability. It creates conditions in which encountering variability is survivable.

The in-between areas in my sketch — the transitional spaces between indoor and outdoor, the semi-covered courtyard where rain is audible but you are not quite wet, the corridor with a window that runs the full height of the wall facing north — these are not compromises. They are the scaffold. The place where children practice attending to the real world before it asks everything of them.

A child who has spent years noticing the pollinator garden will not be alarmed by complexity. They have already learned that living systems have dependencies they didn’t design, timings they don’t control, and gaps that their plans forgot.

They have, in other words, learned to spot edges.

The Upcycling Corner

I want to end with one detail from the outdoor sketch that I keep returning to.

The upcycling corner sits at the far edge of the outdoor space, past the water play and the wildlife garden. It is not the most dramatic element. There is no clever material. No biomimetic principle.

It is simply a space where broken things go, and where children decide what they might become.

The broken thing is honest about its failure. It does not work as it was designed to work. A child who picks it up is confronted immediately with the question: What went wrong? What would it need? What could it be instead?

No technique required. No adult prompt needed. Just a thing that stopped working, waiting for a child who has learned — through years of living with a building that teaches — to ask the right questions.

That is the kindergarten I am trying to sketch.

Not a space that teaches children what to think. A space that teaches them to notice what they don’t know.

Thanks for reading. If this resonates, I’d love to hear what you notice in the environments around the children in your life — what questions those spaces are quietly asking, or quietly preventing.

AI was used as a Thinking Partner

Imagine being a doctor trying to save a patient, but you could only see one-fifth of their body. How would you diagnose, treat, or perform surgery?

This metaphor captures the dilemma we face with Earth’s biodiversity. According to ecologist Sara Beery of MIT, we’re urgently trying to protect life on Earth, yet we only see a fraction of it. Scientists have described roughly 2 million of the estimated 10 million species on Earth, meaning that 80% of biodiversity remains hidden from our understanding.

But what if the knowledge we need is already recorded, pattern upon pattern, in the data we’ve collected? And what if AI could help us see what was always there—just like a telescope reveals distant stars invisible to the naked eye?

Subscribe now

The Ecological Goldmine We’re Sitting On

We stand on a mountain of ecological information.

• iNaturalist hosts over 300 million photos contributed by citizen scientists.

• Xeno-canto preserves millions of animal sound recordings.

• Wildlife Insights and similar repositories hold tens of millions of camera-trap images.

• FathomNet archives thousands of hours of deep-sea video.

Yet, much of this data remains as undiscovered as fossils buried beneath sediment. If you spent just one second on every iNaturalist image, it would take more than 40 years of full-time work to view them all. We are information-rich but insight-poor.

This pattern reflects a deeper ecological truth: life is hugely interconnected and complicated. Just as a rainforest canopy obscures the richness below, sprawling datasets hide layers of knowledge that humans alone cannot excavate — unless aided by tools that see patterns like nature does.

The New Economics of Discovery

AI isn’t replacing human curiosity; it’s extending it.

Sara Beery's INQUIRE system at MIT demonstrates this power. Traditional methods for finding examples of specific bird behaviours—like birds eating insects—required extensive manual data curation that could take researchers months. INQUIRE can accomplish comparable searches in a fraction of the time, transforming research that once consumed entire careers into tasks that can be completed in days or hours.

That’s not a marginal upgrade — it’s a metabolic revolution in scientific discovery, like when photosynthesis evolved and life on Earth suddenly had access to sunlight as fuel. With faster exploration:

• Researchers can follow 10 ideas instead of one.

• Unexpected pathways become visible.

• Creativity expands, not contracts.

This reflects a core ecological strategy: distributed sensing and pattern detection, where many small parts (like leaves on a tree) gather more information together than any single observer ever could.

Seeing What Was Always There

Consider one image labelled Grant’s zebra on iNaturalist.

At first glance, it’s a simple sighting. But look deeper:

• You can identify individual zebras by stripe patterns — foundational for population studies.

• The zebras’ proximity to wildebeest suggests interspecies coexistence.

• An oxpecker on one animal indicates pest control and symbiotic relationships.

• The vegetation in the background can be analysed for biomass and carbon storage.

One photo. Dozens of insights.
Multiply this by 300 million images — and you begin to glimpse how much biological knowledge is hiding in plain sight.

This mirrors how ecosystems work: emergent complexity arises from simple interactions. A single stream doesn’t tell you the shape of the watershed, but map every tributary and suddenly the whole system becomes visible.

AI on the Front Lines of Climate and Biodiversity

AI isn’t a theoretical tool — it’s already operating in real ecological crises:

• AI flood forecasting systems have predicted extreme events days ahead, often in data-poor regions, saving lives and reducing economic harm.

• High-resolution climate models give communities more time to prepare for storms and heatwaves.

• Airborne imagery and machine learning track deforestation and methane emissions — exposing patterns invisible to human analysts.

• AI-enabled wildfire sensors detect the earliest chemical signals of combustion, allowing intervention before fires erupt.

These are not futures — they are present realities.

These tools mirror how nature itself “computes” patterns: dynamic feedback loops where systems sense, respond, and adapt in real time (like how forests regulate fire cycles). AI augments that capability at scales far beyond Unaided human observation.

Cross-Pollination

This isn't just a win for ecology; it’s a blueprint for any complex system. In the same way that "cross-scale interactions" in nature allow a mutation in a single microbe to eventually transform an entire forest, AI-driven discovery is rippling into business and design.

By reducing project timelines from weeks to hours, teams aren't just working faster—they are mimicking an ecological strategy: rapid iteration. Like a polyculture garden that produces more than a monoculture, the "human + AI" partnership generates a higher diversity of ideas than either could alone.

Breaking the Expertise Paradox

There’s a paradox in expertise: the more you know, the more your knowledge can blind you to novel solutions.

AI helps break this constraint.

Machine learning models can identify species traits or habitat signals that scientists have never studied. They illuminate links in ecosystems — like predator-prey networks — that once seemed invisible.

A 2024 horizon scan by conservation scientists identified 21 ways AI could reshape biodiversity work, from discovering “dark diversity” to improving wildlife trade monitoring and conflict response.

This reflects what biologists call distributed cognition — intelligence isn’t just in one brain but emerges from many interacting parts.

The Democracy of Discovery

Perhaps the most transformative effect of AI is who gets to make discoveries.

Millions of citizen scientists now contribute data that — when coupled with AI — becomes part of global knowledge. Small organisations, communities, and individuals can engage in research once exclusive to well-funded labs.

This echoes an ecological principle: diversity increases resilience. More observers, more perspectives, more ways of seeing create a stronger, more adaptive science.

Humans + AI: A Partnership Modelled on Nature

AI doesn’t replace human judgment — it amplifies it.

Scientists still ask the questions. They still formulate hypotheses. They still interpret results.

AI handles the scale — just as ecosystems rely on countless small processes that, together, create vast complexity.

Examples include optimised traffic systems, farm irrigation guided by AI insights, and collaborative lab work where machines accelerate laborious tasks.

This interplay reflects nature’s own strategy: feedback systems where agents act, sense, and adjust — together.

The Resilience of Human Context

Power is not a panacea. AI carries the shadows of its creators: data gaps in the Global South, inherent algorithmic biases, and the carbon cost of compute. These challenges echo a fundamental ecological truth: no system thrives when a single species—or a single data source—dominates.

True resilience comes from "Distributed Stewardship," where we integrate high-tech sensing with local indigenous knowledge and human ethical guardrails. AI gives us the map, but humans must still choose the destination.

A New Treasure Hunt

The biodiversity crisis is real: extinction rates are now 100 to 1,000 times higher than natural background levels. Species face habitat loss, climate shifts, and cascading ecological collapse.

But we also have unprecedented tools. AI reveals patterns in data we already collected, enabling insights that could help protect species before it’s too late.

And like any ecosystem, the whole is greater than the sum of its parts.

Every photo, every recording, every observation contributes to a collective picture. AI helps us see the whole more clearly — but the curiosity, the wonder, and the moral imperative still originate in the human heart.

The Infinite Nature of the Hunt

The hunt for knowledge isn’t finite — and that’s the point.

Unlocking one pattern reveals another. Discovering one species connects to dozens more. Solving one conservation challenge exposes ten more.

This reflects nature’s most fundamental principle: complexity begets more complexity.

AI helps us see deeper into the deep tapestry of life. The treasure was always there — now we finally have the tools to perceive it. And the hunt has only just begun.

Join the Digital Expedition. We are living through the greatest era of discovery in human history. Subscribe to receive weekly dispatches from the intersection of technology and the natural world. Let’s map the "dark diversity" of our planet together.

Subscribe now

Sources

INQUIRE System: arXiv paper 2511.15656 (November 2025)

AI Horizon Scan: Reynolds et al., Trends in Ecology & Evolution (2024)

Extinction Rates: Pimm et al., Science (2014); De Vos et al., Conservation Biology (2014)

Species Estimates: Our World in Data; Mora et al. (2011); multiple Natural History Museum sources

iNaturalist: Wikipedia entry (updated January 2026)

AI was used as a Thinking Partner.

AI-Generated Image.

The word "kindergarten" reminds us that education was once rooted in the soil. A “garden for children” was never just a metaphor—it was a developmental strategy. Before classrooms became rigid, learning unfolded through seasons, textures, weather, and living relationships.

Architecture is a silent teacher: it shapes attention, behaviour, and values without uttering a single word. By 2035, this teacher will no longer stop at the school gate. As the walls between school, home, and world continue to thin, the home must inherit the role of the kindergarten—becoming a daily, lived curriculum in ecological belonging.

From Warehouse to Habitat

Most contemporary homes function as warehouses: storage for people, possessions, and increasingly, captured attention. Screens dominate sensory bandwidth while climate control flattens biological variability. The result is sensory deprivation disguised as convenience.

The 2035 home must undergo a radical transformation: from a warehouse of goods to a habitat for life.

The Living Curriculum

In a true habitat-home, biophilic elements are not decorative; they are legible. Indoor ecosystems—soil, plants, fungi, and water—become mentors. Children learn by noticing the subtle shifts: leaves drooping in thirst, fungi spreading after a rain, or the migration of moisture across a surface.

• The Risk of Metrics: We must avoid “over-instrumenting” nature. If every leaf reports data to a smartphone, we teach children that nature only matters when translated into numbers.

• The Goal: Restorative presence. Interpretation should be offered sparingly—like a field guide, not a narrator.

Biomimetic Play: Lived Physics

The 2035 home replaces branded wonder with material intelligence.

• Adaptive Surfaces: Walls that bead water like a lotus leaf or textures that flex like cartilage allow children to encounter complex physics as a tactile, lived experience.

• The Principle Over the Novelty: The question is not how “clever” a material is, but whether it reveals a biological truth: resilience, efficiency, or reciprocity.

AI as Caretaker, Not Conductor

In 2035, AI will inevitably shape domestic ecosystems, but it must act less like a manager and more like a groundskeeper. This is “tinkering with a purpose”.

1. Translating Without Replacing

Sensors may detect plant stress, but instead of silently correcting the soil, the AI should pause and ask: Who else might help? By inviting the child to observe, hypothesise, and respond, the AI restores the feedback loop between action and consequence.

2. Circadian Belonging

Homes that track solar rhythms—adjusting spectrum, intensity, and shadow—anchor a child’s body to planetary time. However, we should resist perfect replication. A sky that is always “right” teaches predictability, not ecology. A resilient child needs the occasional glare, gloom, and surprise of real weather.

Rethinking the “New Outdoor”

Climate instability in 2035 presents a hard truth: the outdoors is no longer a universal safety zone. Heatwaves and air quality issues will push children inside. But buffering nature is not the same as replacing it.

• Transitional Ecologies: Courtyards, winter gardens, and screened porches act as ecological thresholds. These spaces allow children to feel the wind and hear the storm without being overwhelmed by it.

• The Value of Discomfort: If every environment is curated to perfection, children grow up ecologically fragile. A true habitat includes the lessons of repair and recovery.

Beyond Human-Centred Design: Homes That Practice Empathy

The most important shift is ethical. A house that hosts life—and sometimes fails to meet its needs—teaches responsibility. When a system struggles, the child learns that care is an ongoing verb, not an automatic setting. This is empathy learned through participation.

Perhaps the most radical design choice in 2035 will be what we choose NOT to optimise. Moments of unscripted discovery—a beetle crossing a windowsill or condensation tracing a pattern on glass—must remain free from digital commentary. Wonder needs silence.

The Choice for 2035

The future is not a choice between technology and nature; it is a choice between substitution and relationship.

Will children be raised in homes that simulate life while keeping it safely abstract? Or in homes that behave like ecosystems—messy, instructive, and alive?

By 2035, the home must do more than shelter. It must teach humility, attention, and belonging. Not by telling children what nature is—but by letting them live inside it.

Design not for control. Design for kinship.

AI-Generate Image

AskNature Chat was used for inspiration.

Also, I have written about biophilic design here.

In our last post, we explored how AI accelerates our ability to harvest nature’s genius. But speed without direction is just a faster way to get lost. If nature is truly our mentor, then our inventions must be more than clever—they must be respectful.

Ethics in biomimicry isn’t a boundary that holds us back; it is the “humus”—the rich, organic foundation that allows an invention to truly take root and thrive.

Here is how we move from learning about life to inventing with it.

1. The Principle of Rooted Purpose

Nature never innovates for the sake of novelty. A mutation only persists if it serves the fitness of the whole. Before engaging an AI to “optimise” a biological pattern, we must find our own “ecological niche.”

• The Question: Are we solving a human vanity, or a planetary need?

• The Practice: Align your design brief with life’s laboratory. Don’t just ask AI for “faster” or “stronger”; ask for “integrated” and “resilient.”

2. Radical Reciprocity - The Student’s Tithe

When we learn from a coral reef or a fungal network, we are taking a gift of 3.8 billion years of R&D. In nature, no transaction is one-way.

• The Practice: If your material is inspired by a threatened species, your business model should support its restoration. Partner with conservationists as “technical consultants.” Treat the ecosystem as a stakeholder with a seat at the board table.

3. Honouring the Ghost in the Data

AI thrives on abstraction—stripping the “pattern” from the “place.” But a bird’s wing is inseparable from the wind currents of its home.

• The Practice: Use AI to model context, not just form. Don’t just ask what a strategy is; ask what it costs the organism in energy, and what trade-offs it made to survive. Patterns without context are like seeds without soil—they have potential, but no soul.

4. More-Than-Human Metrics

Traditional engineering measures cost, weight, and speed. Life measures affinity.

• The Practice: Expand your dashboard. If your AI-optimised drone is 10% more efficient but its manufacturing poisons the watershed of the very bird it mimics, the design is a failure. We must measure our “ecological footprint” as carefully as our “profit margin.”

5. Designing for the Great Return

In nature, there is no “away.” Every endpoint is a beginning for something else.

• The Practice: Use AI to simulate the “afterlife” of your invention. How does it decay? Does it leave the soil better than it found it? Use predictive modelling to ensure your invention can gracefully re-enter the biological cycle.

6. The Digital Footprint: Respecting the Silence

The “clouds” where our AI lives are cooled by real water and powered by real earth. An ethical biomimicry inventor recognises that the tool itself is part of the planet's metabolism.

• The Practice: Practice “computational temperance.” Choose energy-efficient models. Don’t run a thousand simulations if ten thoughtful ones will do. A tool inspired by life should not consume the conditions that make life possible.

The Integrity Check

Before a seed becomes a sapling, it must survive the environmental conditions. Before your invention scales, put it through the Living Test:

• Does it nourish? Does this design give back to the system it learned from?

• Is it honest? Does it acknowledge its biological “ancestors” and its digital costs?

• Is it fit? Does it solve a problem for the whole, or just for the part?

• Is it humble? Does it leave room for the ecosystem to change and breathe?

Seeds for the Field

• The Easiest Step: Which part of this covenant feels like “common sense” to your current practice?

• The Hardest Climb: Why does “Reciprocity” (giving back to the ecosystem) feel so difficult in our current economic climate? How can we hack that system?

• The Ghost Species: Have you ever been inspired by a creature that is vanishing? How does that change the “weight” of your patent?

• The Memory of Machines: What kind of ancestor do you want to be? When future generations (human and otherwise) look back at your work, will they see a partner or a parasite?

AI-Generated Image

AskNature Chat and Google Gemini were used as inspiration.

For the first time in our history, artificial intelligence allows us to reach into the deep well of evolutionary time and pull up secrets in seconds. We are no longer just observing nature; we are using digital “mycelium”—vast, invisible networks of computation—to scan, model, and recombine biological strategies at speeds evolution never intended.

The architecture of a leaf’s veins becomes a flow algorithm for our cities. The molecular dance of spider silk becomes code for our factories. The cooling breath of a termite mound becomes a climate model for our skyscrapers. For the biomimetic inventor, this is a threshold moment: a promise of infinite insight and the end of the “dead end.”

But beneath this exhilaration, a quiet, tectonic tension shifts: What happens when we learn from life faster than we learn how to live within it?

Can Wisdom Survive Abstraction?

Nature has never taught through data points; it teaches through relationships. Biomimicry was born not as a technical shortcut, but as a homecoming—a shift from seeing nature as a warehouse of parts to seeing it as a council of mentors.

A forest does not thrive by maximising a single output. It survives by balancing the tension between growth and decay, competition and radical cooperation. AI, however, excels at extraction. It strips the pattern from the soil. It lifts the “genius” of an organism out of its ecosystem so it can be applied to a human problem.

When a biological strategy is severed from its constraints, its predators, and its neighbours, it loses its “ecological weight.” It may remain clever, but it ceases to be wise. As inventors, we must ask: Are we translating nature’s language, or just harvesting its vocabulary to tell our own stories?

The Friction of Time: Speed vs. Stability

In the living world, speed is rarely the ultimate goal. Persistence is.

A coral reef takes centuries to craft its calcified cathedral. Soil requires a slow, rhythmic patience to regenerate. Life tests ideas gently, over aeons, through a relentless cycle of feedback.

AI rewards the opposite: acceleration, scale, and the optimisation of variables. This creates an ethical paradox. We turn to nature because we crave its resilience, yet we use tools that reward fragility by demanding instant results. If we use AI to accelerate “green” innovation without respecting the slow rhythms of the Earth, we risk building a world that is efficient in the short term but ecologically hollow in the long term.

From Extraction to Reciprocity

Biomimicry has always walked a knife’s edge between learning and extraction. AI sharpens that edge.

When we turn organisms into “data sources,” we risk slipping back into the old, colonial mindset of the Industrial Revolution—treating the biosphere as a library to be looted rather than a family to be protected.

The ethical question is no longer just what we can learn, but how we give back. What does it mean to design a “reef-inspired” material while the reefs themselves are bleaching into silence? A true student does not just take the lesson; they ensure the teacher survives to teach again.

The Question Beneath the Question

The challenge of our era is not technical; it is a matter of the heart.

• AI asks: “What works?”

• Nature asks: “What lasts?”

As inventors standing at the interface of biology and bits, we are the bridge-builders. Our responsibility is to ensure that our tools do not just mimic the forms of life, but honour the logic of life. We must move beyond “bio-inspired” to “life-aligned.”

Seeds for Reflection

• The Line of Extraction: Where do you draw the line between being a student of nature and being a harvester of its data?

• The Loss of Context: When a bird’s wing becomes a mathematical curve in an AI model, what essential “truth” about the bird is left behind?

• The Gift of Slow: Is there such a thing as “slow AI”? How can we use fast tools to support slow, enduring ecological health?

• The Debt of Invention: If an organism inspires a profitable patent, what does the inventor owe to that organism’s habitat? What would “technological reciprocity” look like?

In our next post, we will move from the forest floor to the laboratory—exploring how we can build frameworks for "Regenerative AI" that prioritise the whole over the part.

AI-Generated Image

AskNature Chat and Google Gemini were used for research and inspiration.

Further Reading & Inspiration

The Science of the Patterns

• On Leaf Veins & Urban Flow: The Plant Leaf: A Biomimetic Resource for Multifunctional and Economic Design, MDPI (2023). This study explores how the hierarchical branching of leaves provides a blueprint for resilient infrastructure.

• On Spider Silk as Code: Correlating Mechanical Properties and Sequence Motifs in Artificial Spider Silk, ACS Biomaterials (2024). A deep dive into how we are decoding silk proteins into programmable digital motifs.

• On Termite Architecture: Termite Mound Architecture and Climate Control, Royal Society Publishing (2025). This review uses X-ray tomography to show how mounds “breathe,” providing a masterclass in passive cooling for modern skyscrapers.

The Intersection of AI & Biology

• On AI Optimization: From Nature to Neural Networks: AI-Driven Biomimetic Optimization, MDPI (2025). An exploration of how AI parameterizes biological patterns to solve complex engineering hurdles.

• On Digital-Biological Cooperation: Towards Beneficial AI: A Biomimicry Framework, MDPI (2025). A foundational paper on designing intelligence that supports, rather than replaces, biological entities.

The Ethical Foundation

• Biomimicry Institute (AskNature.org): The essential resource for the “Design Lens” framework, emphasizing Nature as Model, Measure, and Mentor.

• The Environmental Footprint of AI: The Environmental Footprint of Artificial Intelligence, arXiv (2025). A critical look at the physical resources required to run the models that learn from life.

It does not encounter it in terms of tensile strength, elasticity, or weight-to-performance ratio. For the spider, silk is tension and response — the faint tremor of an insect’s wing, the warning of a breeze, the architecture that holds eggs, memory, and safety in place. A web is not a thing the spider makes. It is a conversation the spider holds with its world.

When we lift spider silk out of that web, isolate its properties, and call that innovation, we are only partly right. We gain utility. But we leave behind context — the dense weave of relationships that gave that silk meaning in the first place. In doing so, we mirror a deeper pattern in how we engage with the living world: a habit of taking what works, while setting aside the relationships that made it possible.

Biomimicry, as it is often practised today, can feel like a walk through a well-stocked hardware shop. We move through nature’s aisles in search of useful components: a stronger fibre, a cleaner surface, a more efficient structure. Spider silk becomes a materials problem. Lotus leaves become a coating. Shark skin becomes a texture. These innovations are not wrong. Many are elegant, even necessary.

But they are also reductive. They reflect a broader cultural attitude towards nature — one that values what can be extracted, measured, and owned, while quietly disregarding the ecological context that produced it.

What gets lost is not information, but the relationship.

What Gets Lost When We Pull Things Apart

When we extract a function from a living system, we often congratulate ourselves on clarity. Complexity has been simplified. Something wild has been made legible. Yet in ecosystems, clarity is often a form of amputation.

Take fungi. In many biomimetic conversations, mycelium is reduced to what it does — decomposition, soil formation, carbon cycling. These functions are real, but they are only part of the story. Left intact, mycelial networks are not merely chemical processors. They are living mediators — weaving together trees, microbes, minerals, water, and time into an adaptive whole.

Beneath forest floors, mycelium redistributes resources according to need and opportunity. It ferries sugars from sun-rich trees to shaded seedlings, trades minerals gathered from soil for carbon fixed in leaves, and adjusts its growth in response to drought, disturbance, and competition. It does not optimise a single outcome. It maintains relationships.

When we extract fungi from this context — turning mycelium into a material, an algorithm, or a metaphor — we gain something useful. But we lose the very quality that makes it powerful: its responsiveness to the living system it inhabits.

The same narrowing happens when forests are framed primarily as carbon sinks. Carbon storage matters. But when carbon becomes the forest’s defining feature, ancient ecosystems and monoculture plantations begin to look interchangeable on a spreadsheet. Biodiversity becomes a side benefit. Water cycling, soil formation, cultural meaning, and long-evolved interdependence fade from view.

The solution appears to work, yet the system quietly unravels.

This is not a failure of intent. It is a failure of framing.

A Different Kind of Question

Relational biomimicry begins by asking a different kind of question.

Not What does this do?
But What does this participate in?

• What does the spider’s web participate in?

• What does mycelium hold together?

• What does a forest become over centuries, rather than years?

When we ask these questions, different patterns come into focus. We see structures that serve many functions at once. Systems that trade speed for resilience, efficiency for adaptability. Organisms that survive not by optimising a single trait, but by fitting well within a web of relationships.

This way of learning is slower. It resists shortcuts. It asks us to stay with complexity rather than strip it away. And it often leads us away from neat, universal solutions towards designs that are local, layered, and responsive. Which is precisely why it matters.

What Relationship-Based Biomimicry Might Look Like

A relational approach to biomimicry would not abandon engineering or materials science. It would situate them. Designs would be evaluated not only by how they perform in isolation, but by how they behave once embedded in living systems.

Inspired by mycelial networks, we might imagine human systems that redistribute resources dynamically rather than locking them into rigid channels; that reward long-term contribution rather than short-term extraction; that tolerate redundancy instead of pursuing brittle efficiency; that adapt their structure to place rather than imposing uniform solutions everywhere.

From ecological succession, we might learn to design systems that prepare the ground for what comes next — valuing early, imperfect stages not for their permanence, but for how they enable future complexity. From symbiosis, we might rediscover collaboration not as a transaction, but as a deep partnership — the kind that generates capabilities no single participant could achieve alone.

These are not metaphors. They are operating principles in living systems.

The Cultural Mirror

Seen clearly, what this approach to biomimicry reveals is not just a design problem, but a cultural one.

Our technologies reflect how we think. If we are trained to value parts over wholes, speed over patience, ownership over stewardship, then our nature-inspired designs will quietly reproduce those values — even when they are labelled sustainable.

When we extract mechanisms without honouring relationships, we imply that the part is valuable, but the web that produced it is optional. That assumption does not come from nature. It comes from us.

Nature invests relentlessly in relationships. Nothing stands alone for long.

The Harder Lesson

Which leads to an uncomfortable possibility.

• What if the most important lessons nature has to teach us cannot be patented

• What if they cannot be isolated in a lab or scaled without context?

• What if they only make sense when practised in place, over time, within living systems?

What if the wisdom is not in the spider’s silk, but in the spider’s way of being a spider — attentive, embedded, responsive, inseparable from the web it inhabits?

That kind of learning is harder to monetise. It asks more of us. It requires humility, patience, and a willingness to be changed by what we study.

But it may be exactly the kind of learning this moment calls for.

Because biomimicry, at its deepest, is not just about learning from nature.
It is about learning how to belong — to place, to each other, and to the living systems that make any future possible.

Closing Reflection

As we look towards 2026, perhaps the invitation is not to extract better ideas from nature, but to listen more carefully to the relationships that sustain life. To design with restraint as well as imagination. To measure success not only by what we build, but by what we help endure.

May the coming year bring patient curiosity, innovation that is rooted, and a renewed sense of belonging within the larger web we are already part of.

Here’s to an inspiring, relational, and life-affirming 2026.

AskNature Chat was used for inspiration.

Yet philosophy alone does not move societies.

We can intellectually grasp that a “Satoyama city” is possible, but we cannot see it. We cannot walk its streets, feel its shade, or hear birdsong returning to its corners. And as climate scientist Simon Donner has argued, this failure of imagination is not cosmetic. It is structural.

This essay makes a modest, but demanding claim: climate action has stalled not because we lack data, nor because people are incapable of imagining better lives, but because our collective imagination has been systematically severed from the places where political struggle actually occurs.

Imagination does not replace power. But without imagination, power is rarely confronted with a coherent demand.

The Problem Is Not Ignorance, but Distance

Simon’s work describes climate change as suffering from psychological distance. It feels like something happening elsewhere, to other people, at another time. We can model a 2°C rise with extraordinary precision, yet remain strangely unable to picture what a flourishing post‑carbon neighbourhood might actually feel like.

This does not mean imagination is the primary constraint. Fossil fuel incumbency, capital lock‑in, zoning regimes, labour precarity, and political capture remain the dominant obstacles. No image, however compelling, dissolves entrenched power on its own. But movements rarely challenge power without first articulating what they are fighting for. Imagination does not substitute for struggle; it gives direction to struggle.

If we cannot imagine the alternative, we are left only managing decline.

World Models as Scaffolding, Not Solutions

This raises a careful question: can AI—specifically the emerging class of systems known as world models, such as Genie 3—help repair this imaginative rupture?

World models are still partial and imperfect, for example, Genie 3 has not been launched to the public yet. They attempt to simulate aspects of physical reality: how light interacts with surfaces, how water moves through cities, how landscapes respond over time. Unlike language models that predict text, these systems attempt to model behaviour. They do not understand the meaning. They do not possess values. And they are not neutral.

These systems are products: trained on uneven data, shaped by existing aesthetic norms, owned by institutions with their own incentives. Left unchecked, they will reproduce the same affluent, sanitised futures familiar from developer renderings and sustainability brochures.

For this reason, world models must never be treated as authorities. At best, they function as scaffolding: provisional structures that support collective imagination while being continually revised, contested, and—when no longer needed—dismantled.

Used this way, AI becomes neither saviour nor solution, but a lens through which latent possibilities can be made visible again.

The promise is real. The danger is equally real.

Ending Abstraction by Returning to Place

Hyper‑local futures

Psychological distance thrives on generality. Global averages and distant catastrophes fail to touch daily life.

A world model, by contrast, can be anchored to a specific place: your street, your bus stop, or your local park. It can show that place under climate stress—or reimagined as a productive edge between human and wild.

When the future is rendered in abstraction, it remains remote. When it is rendered in place, it becomes negotiable. The impossible shifts into the realm of the unfinished.

Beyond sacrifice narratives

Climate action is still too often framed as deprivation: what must be given up, restricted, or endured. In this framing, mottainai becomes moral scolding rather than ethical care.

Grounded simulations can reveal something else: the co‑benefits that rarely appear in policy spreadsheets. Quieter streets. Cooler summer walks. The return of birdsong masked by traffic noise. Infrastructure repaired rather than replaced, bearing visible traces of care rather than erasure.

There is a danger here. Beauty can pacify as easily as it can mobilise. A lush simulation risks becoming a consolation prize—something people admire virtually while their real neighbourhoods remain polluted and neglected.

For this reason, the goal is not serenity. It is about desirability under constraint. Futures that look worth fighting for, precisely because they appear incomplete, contingent, and shaped by political choice.

The rehearsal space

Interactive models allow societies to rehearse change. What if this car park were partially reclaimed? What if this office block were retrofitted rather than demolished? What if this median strip became habitat?

For policymakers, this offers a way to explore risk without pretending it disappears. For residents, it provides a shared reference point that can be carried into meetings, hearings, and campaigns.

The future stops being an abstraction and becomes an argument.

The Limits Must Be Named

Any use of simulation that ignores power will fail. Visualisation does not dismantle fossil capital. It does not rewrite zoning codes. It does not redistribute labour or land. There is a genuine risk that imaginative work diverts energy from the slower, harder work of organising and confrontation.

History offers ample evidence of this. People rarely act because they possess a clear image of a better future. More often, clearer visions emerge because people have begun to act.

This reverses the usual narrative. Imagination cannot come first. It must be embedded within struggle.

From Engineering to Gardening

Engineering asks: How do we optimise?

Gardening asks: What wants to grow here?

If AI is used purely to optimise carbon reduction, it repeats the mechanistic logic that helped create the crisis. Efficient systems can still feel dead.

Used differently, world models resemble a gardener’s almanac: not a plan, but a guide. A way of noticing patterns, edges, and seasonal possibilities.

Data tells you soil chemistry and frost dates. Imagination is what makes you plant the garden at all. Climate science shows where the rocks are. It does not steer the ship.

Why Imagination Matters

Simon bridges physical climate science and human perception. He understands that climate change is not only a physics problem, but a problem of time, place, and narrative.

We possess extraordinarily detailed models of what will happen if we do nothing. We have almost no shared models of what else could happen if we act.

This asymmetry matters.

By combining the ethical depth of ikigai, mottainai, and satoyama with carefully constrained AI simulations, we can move from knowing the problem to seeing a defensible alternative. Not a utopia. A direction.

The Imagination Lab: Re-mapping the “Grey”

We’ve spent too long looking at maps of what we might lose. It’s time to start looking at maps of what we can build.

Simon reminds us that the biggest obstacle to climate action is our inability to see the alternative. If we can’t visualise a concrete city centre as a living, breathing Satoyama mosaic, we will never summon the political or personal will to create it.

I want to help bridge that gap with you.

I am launching the “Imagination Lab”—a small project where we use the principles of Ikigai, Mottainai, and Satoyama alongside AI World Models to re-imagine the specific places you live and work.

How to Participate

1. Identify the grey
   A car‑dominated junction, a lifeless plaza, a logistics corridor, a neglected estate.

2. Describe it plainly
   Where is it? What does it currently optimise for? What feels absent or imposed?

3. Collective re‑imagining
   Selected submissions will be explored using the framework below, producing simulations intended for real‑world advocacy.

When you attend a council meeting, you should be able to say:

This is what we are asking for.

Why take part

This isn’t just a digital art project. It is an exercise in agency. By localising the future, we shrink the “psychological distance”. When you can see your street transformed from a heat-island into a sanctuary of Ikigai, the climate crisis stops being a distant data point and starts being a creative invitation.

Leave a comment below with a “Grey Space” you want to see transformed. Let’s see what happens when we stop being engineers and start being gardeners of the future.

Its purpose is not to produce artworks, but arguments: visual reference points that communities can use in planning disputes, budget negotiations, and public debate.

You do not need perfect language. You do not need optimism. Anger, fatigue, and grief are welcome inputs.

The Imagination Lab Framework

Pillar 1: The Satoyama Mosaic

Rule: No monocultures
Create diversity of species, heights, and functions. The boundary between built and wild should be porous, not decorative.

Pillar 2: Mottainai and Kintsugi

Rule: No wholesale replacement
Transform through repair. Show history. Honour embodied carbon by letting life emerge through cracks rather than erasing them

Pillar 3: Ikigai Space

Rule: No optimisation‑only design
Include space for rest, contemplation, or joy that serves no economic function

The Temporal Rule

The future must appear mid‑transition. Some elements are young. Some repairs are rough. The work is visibly ongoing.

The Filter

Every simulation must be grounded in physical reality.

Apply thermal modelling consistent with mid‑century projections. Show reduced heat through shade and evapotranspiration. Let light, glare, and air movement convey cooling without pretending the climate problem vanishes.

Making the Invisible Contestable

The world we want does not yet exist—but it already occupies the realm of possibility.

It lives in the productive edges between what we have built and what we have forgotten how to imagine. The task is not to escape politics through vision, but to return to politics with something concrete in hand.

What grey space in your city is ready to be argued over?

Testing the Framework: Two Cities, Two Futures

I tested this approach in two starkly different contexts: New York City and N’Djamena, Chad.

New York was chosen because it embodies the mechanistic urbanism we’ve been discussing—a city I know firsthand, where summer concrete radiates stored heat long after sunset, where glass towers amplify sun into weapons, where the hum of a million air conditioners drowns out any possibility of birdsong. It’s also a major contributor to the crisis: a wealthy metropolis whose infrastructure and consumption patterns have written the climate problem in carbon.

N’Djamena represents the inverse injustice. Chad has contributed almost nothing to global emissions—less than 0.01% of the cumulative total—yet faces some of the most severe climate impacts on Earth. Lake Chad has shrunk by 90% since the 1960s. Temperatures are rising faster than the global average. The capital city, built for a different climate, now faces extreme heat that makes daily life increasingly unbearable.

If our Imagination Lab is to mean anything, it cannot only reimagine wealthy cities that caused the problem. It must also offer vision to places bearing consequences they didn’t create—places where the “failure of imagination” isn’t a luxury problem but a survival question.

The Manhattan Canyonlands (Year 2050)

Viewpoint: Standing at the intersection of 5th Avenue and 42nd Street, looking south on a mid-July afternoon. Outside the simulation, the temperature is a blistering 98°F (36.6°C).

1. The Structural Transformation (Mottainai & Kintsugi)

The World Model respects the embodied carbon of the existing skyscrapers. Nothing has been demolished.

• The Glass Skins: The iconic glass curtains of the towers remain but are retrofitted with an external “second skin” of reclaimed aluminium latticework. This shading infrastructure reduces solar gain by 40%, slashing the energy needed for air conditioning.

• The Street as Sponge: The endless expanse of concrete road has been “broken.” Following the Kintsugi aesthetic, fissures have been intentionally widened and filled with a permeable resin made from recycled plastics and local aggregate. These golden-hued cracks channel intense rainfall into subterranean aquifers beneath the street, preventing flash floods and nourishing the new street-level ecology.

2. The Ecological Overlay (Satoyama Mosaic)

The model applies the Satoyama principle: maximising the “productive edge” between human habitat and nature.

• Vertical Wetlands: The spaces between the new latticework and the old glass are no longer empty air. They are cascading hydroponic terraces. You see a mosaic of drought-resistant berry bushes, medicinal herbs, and native pollinator-friendly flowers ascending 60 stories. The buildings are literally sweating—transpiring moisture that cools the canyon air.

• The Canyon Floor: The six-lane avenue is reduced to two lanes for essential electric transit. The reclaimed space is a chaotic, dense mix of hardy urban forestry—Ginkgo trees, known for their resilience, form a dense canopy that shades the sidewalk completely.

3. The Climate Foresight Data Layer

You toggle on the “Thermo-Simulation” layer in the World Model.

• The Cool Pocket: The simulation shows a heat map. While the ambient temperature of the city is nearly 100°F, inside this “Satoyama Canyon,” the temperature is a manageable 86°F (30°C). The combination of shade, plant transpiration, and permeable ground has created a localised microclimate of survival.

• The Soundscape: The roar of traffic and massive AC units is gone. The dominant sounds in the simulation are the rush of water moving through the vertical irrigation systems and the chirping of countless insects and birds that have migrated back into the city centre.

4. The Human Element (Ikigai)

The model ensures efficiency isn’t the only metric.

• The “Pause” Points: Where a bank entrance used to be, there is now a small, shaded alcove with benches carved from the wood of trees lost in previous storms. It serves no commercial purpose. It is simply a place for an office worker to sit for ten minutes and watch a Monarch butterfly—a species once rare in NYC—land on a milkweed plant growing out of a 40th-story ledge. It is a space designed purely for the ikigai of connection.

The Verdict of the Simulation

The World Model shows that the concrete is not the enemy; it is the substrate. By applying Mottainai to retrofit rather than replace, and Satoyama to treat every vertical surface as a potential ecosystem, Manhattan transforms from a heat-generating machine into a cooling, productive landscape.

The failure of imagination is cured not by seeing a utopia, but by seeing a viable, beautiful adaptation of what already exists.

AI-generated image.

The Dust Silence (Chad, 2050)

Viewpoint: The outskirts of N’Djamena, moving toward the remains of the Lake Chad basin. Ambient Conditions: The “Wet Bulb” temperature frequently exceeds 35°C, the limit of human survivability. The air is thick with fine particulate dust from the encroaching Sahara.

1. The Death of the Oasis (Failure of Satoyama)

In this scenario, the “productive edge” has been obliterated.

• The Vanishing Lake: Lake Chad, once a source of life for 30 million people, has retreated into disconnected, stagnant seasonal puddles. The simulation shows a “Resource Conflict” overlay—red zones where remaining water sources are militarised.

• Ecological Collapse: The mosaic of savannah and scrubland has been replaced by a monolithic “dust-scape.” Without managed intervention or regenerative grazing, the soil has lost all biological structure, becoming hydrophobic. When it does rain, the water doesn’t sink in; it causes flash floods that wash away the remaining topsoil.

2. The Ruin of Infrastructure (Failure of Mottainai)

Instead of the “Sacred Repair” of Kintsugi, we see the exhaustion of materials.

• The Abandonment: The simulation shows vast areas of rural Chad marked as “Ghost Zones.” Abandoned mud-brick and concrete structures are half-buried in sand dunes. This is the opposite of Mottainai; these materials weren’t honoured or repurposed—they were simply outpaced by a changing climate and left behind.

• The Energy Gap: Without a transition to decentralised green grids, the old fossil-fuel infrastructure has crumbled under the weight of extreme heat and supply chain collapses. The simulation shows a silent landscape, devoid of the mechanical hum of a functioning modern state.

3. The Loss of Meaning (Failure of Ikigai)

This is the most profound part of the simulation.

• The Migration Pulse: The “Human Layer” of the model shows a massive, forced northern and southern movement. Ikigai—the joy of being and the purpose of place—is lost as millions become climate refugees.

• The Cultural Erasure: When a landscape becomes uninhabitable, the traditions tied to that land (the fishing songs of the Buduma people, the nomadic routes of the Toubou) vanish. The World Model registers this as a “High Loss of Intangible Heritage.”

4. The Reality Check

The “Thermo-Simulation” toggle shows:

• Albedo Feedback: As the dark vegetation dies and is replaced by pale sand, the ground reflects more heat, but the lack of moisture means there is no “evaporative cooling.”

• The Heat Island effect: N’Djamena becomes a “death trap” of trapped solar radiation, with no green corridors to provide relief.

Why this Visualisation Matters for Action

If we “do nothing,” the World Model shows that Chad doesn’t just get “hotter”—it undergoes a phase shift from a living culture to a geographic casualty.

However, by visualising the specific mechanisms of failure (soil death, water conflict, loss of heritage), we can identify the specific levers of change (regenerative agriculture, water sharing treaties, decentralised solar) that can prevent this timeline.

Here we can see the power of choice.

AI-generated image.

AI-generated image.

AI-generated Infograph

AI was used as inspiration.

When we talk about climate change, we reach for the language of engineering: carbon capture, efficiency metrics, net-zero targets and optimisation. We measure forests in tons of sequestered carbon, oceans in parts per million of acidity, and species in ecosystem services rendered. We have become so fluent in the vocabulary of mechanisms that we’ve forgotten how to speak about meaning.

But what if the problem isn’t that we lack the right technology? What if the problem is that we’re asking the wrong questions?

Three concepts from Japanese philosophy offer a radically different starting point—not as exotic wisdom to be appropriated, but as challenges to the mechanistic thinking that got us into this crisis in the first place.

Ikigai: Why Measurement Fails Us

Ikigai is often mistranslated as “purpose in life” and reduced to a tidy Venn diagram—the intersection of what you love, what you’re good at, what the world needs, and what you can be paid for. But this corporate repackaging misses the point entirely.

Real ikigai resists evaluation. Watching butterflies during your morning run. Tending flowers in your garden. Dancing badly but joyfully. These activities generate no profit, solve no global problem, demonstrate no particular skill—and yet they are central to what makes life worth living.

This matters for climate work because we’ve made the same mistake with nature. We’ve reduced forests to carbon sinks, wetlands to flood control infrastructure, and pollinators to agricultural inputs. We’ve optimised the living world into a spreadsheet, and wonder why our “solutions” feel hollow, why people burn out, and why technical fixes never quite address the deeper malaise.

Ikigai reminds us that not everything valuable can be measured. That joy, beauty, and meaning aren’t inefficiencies to be eliminated—they’re the point. That a thriving planet isn’t one where every acre is optimised for maximum utility, but one where there’s room for the immeasurable, the inefficient, the simply wonderful.

Satoyama: The Productive Edge

If ikigai asks us to honour what can’t be measured, satoyama shows us how to design systems that embrace complexity rather than simplifying it away.

Satoyama refers to the border zone between mountain foothills and arable flatland in Japan—but it’s more than geography. For centuries, these landscapes were actively managed as mosaics of mixed forests, rice paddies, grasslands, streams, and ponds. The result wasn’t wilderness (though it looked wild) or farmland (though it was productive). It was something richer: a living edge where human needs and biodiversity flourished together.

Here’s what makes satoyama revolutionary for biomimicry:

Nature loves edges. The satoyama mosaic created countless transition zones where different habitats met, and these edges teemed with life. Dragonfly naiads grew in spring paddies. Coppiced forests provided fuel while maintaining habitats for butterflies and fireflies. Species didn’t thrive despite human activity—many thrived because of the specific disturbance patterns humans created.

Partnership, not preservation. The Western environmental movement often frames humans as the problem—get us out of the way, and nature will recover. Satoyama suggests the opposite: many species evolved alongside human management patterns and now depend on them. When management ceased in modernising Japan, forests shifted from diverse deciduous stands to monotonous evergreens, and over half the species on Japan’s Red List found themselves without habitat.

Diversity through small-scale variation. Satoyama landscapes weren’t designed by central planners with optimisation models. They emerged from thousands of small decisions by farmers and foresters responding to local conditions. The diversity of the whole arose from the autonomy of the parts—a lesson our age of industrial monoculture desperately needs to relearn.

The tragedy is that satoyama landscapes are disappearing, caught between abandonment (as young people leave for cities) and intensification (as remaining farms industrialise). The same mosaic thinking that created extraordinary biodiversity is being replaced by the binary thinking that asks: wilderness or farmland? Efficiency or waste? Human or nature?

Mottainai: The Spirituality of Not Wasting

Mottainai is often translated as “what a waste,” but this captures only the surface. The word carries a weight of regret, even grief—a sense that when we waste something, we’ve broken a relationship.

Its etymology reveals deeper layers: the term combines mottai—referring to the inherent essence or sacred worth of things in Buddhist philosophy—with nai, a negation meaning “lacking” or “without.” To waste something is to negate its sacred worth, to fail to honour its inherent dignity. And because nothing exists independently in Buddhist thought—everything comes into being through relationship with everything else—to waste food isn’t just economically inefficient, it’s a rupture in the web of connections that brought that food to your plate. The rain, the soil, the farmer’s labour, the life of the plant or animal itself.

This isn’t about guilt. It’s about recognition.

In traditional Japanese practice, you say itadakimasu before eating—a phrase that acknowledges the plants, animals, farmers, and countless others who made your meal possible. Every part of an ingredient is used: fish bones for broth, vegetable scraps for pickling. Broken ceramics aren’t discarded but repaired with gold-dusted lacquer (kintsugi), transforming cracks into gleaming patterns that honour an object’s history rather than hide it.

For climate action, mottainai offers what technical solutions can’t: a reason to care that goes deeper than optimisation.

We already know we waste too much. We have the data on food waste, planned obsolescence, and the fossil fuels we burn. What we lack isn’t information—it’s the felt sense that waste is a form of violence against a relationship. That every discarded object, every squandered resource, every extinct species represents not just lost utility but severed connection.

Mottainai suggests that sustainability isn’t primarily a technical challenge but a spiritual one. Not “how do we optimise resource use?” but “how do we remember that everything we touch is in relationship with us?”

The irony cuts deep: despite this philosophy, Japan discards nearly 5 million tons of edible food yearly (figure from 2023 - 4.64 million tons). Cultural wisdom doesn’t automatically translate to collective action. But perhaps that’s the point—mottainai isn’t a solution to import, it’s a question to sit with: What would it mean to build a civilisation that treats waste as sacred regret rather than economic inefficiency?

The Synthesis: A Different Kind of Climate Thinking

Together, these three concepts sketch an alternative to the mechanistic thinking that dominates climate discourse:

Ikigai reminds us that not everything valuable is measurable—that joy, beauty, and meaning aren’t luxuries to pursue after we’ve solved the “real” problems, but the foundation of any solution worth having.

Satoyama shows us how to design living systems that embrace complexity—where human needs and biodiversity don’t compete but co-create richness through partnership and mosaic thinking.

Mottainai grounds environmental action in relationship and reverence rather than metrics and moralism—transforming sustainability from an obligation into an expression of interconnection.

This isn’t about romanticising Japanese culture or importing solutions wholesale. Japan faces its own environmental crises, its own failures to live its wisdom. Rather, these concepts are provocations—they ask us to examine the assumptions we’ve stopped noticing.

What This Substack Will Explore

In the coming blog posts, I’ll explore:

Biomimicry beyond mechanisms: How can we learn from nature’s designs without reducing them to mere engineering templates? What happens when we copy the spider’s silk but ignore the web of relationships that make spinning silk meaningful? This means I will explore and ask questions such as: What would biomimicry look like if we learned from nature’s relationships and patterns of organisation, not just its technical solutions?

Biophilia as necessity, not luxury: Why contact with nature isn’t a middle-class indulgence but a fundamental human need—and how climate solutions that ignore this are doomed to fail or produce new forms of suffering.

The mosaic principle: What satoyama teaches us about designing landscapes, cities, and economies that thrive on diversity rather than uniformity—and why small-scale variation might be more powerful than grand unified plans.

Meaning in a heating world: How do we maintain ikigai—the joy of being alive—while facing climate grief? Can we hold both the urgency of action and the necessity of beauty?

Sacred materialism: How mottainai and other wisdom traditions challenge the false choice between spirituality and practical action, suggesting that the deepest environmentalism is neither religious nor secular but something stranger and more necessary.

This isn’t about abandoning data, technology, or urgent action. We need carbon budgets and renewable energy, and policy changes. But we also need to remember that we’re not trying to optimise a machine—we’re trying to live well on a living planet.

The question isn’t just “How do we reduce emissions?” but “What kind of humans do we want to become?” Not just “How do we preserve biodiversity?” but “What does it mean to live in a real relationship with other beings?”

These questions can’t be answered with metrics alone. They require wisdom traditions, aesthetic sensibilities, and moral imaginations. They require us to think less like engineers and more like gardeners—or perhaps like the farmers who once tended satoyama landscapes, working with complexity rather than simplifying it away.

Welcome to this exploration. Let’s see what we find at the edges.

If these ideas resonate with you, I’d love to hear your thoughts. What wisdom traditions or practices have shaped how you think about nature and climate? What questions are you sitting with? Reply or leave a comment—this is a conversation, not a lecture.

AI-Generated Image.

AI was used for research.