Get Your Weekly Digest

Receive the best podcast summaries and insights delivered to your inbox every week.

We respect your privacy. Unsubscribe anytime.

TED Talks Daily

The army of autonomous robots restoring nature | Tom Chi

with Tom Chi

Published November 21, 2025
View Show Notes

About This Episode

Impact investor Tom Chi challenges the popular belief that economic growth must come at the expense of nature, arguing instead that the economy is physically a subset of the ecology because everything is ultimately mined or grown. He quantifies the scale of current extraction and describes how outdated industrial processes damage ecosystems, then presents three key shifts: closing material loops through advanced recycling, transforming agriculture with regenerative practices and AI-guided breeding, and using robotics for large-scale restoration on land and underwater. Through concrete examples-from battery recycling and adaptive crops to mangrove-planting drones and a low-cost coral and seagrass-planting robot-he illustrates how modern technology can actively repair ecosystems while supporting a resilient future economy.

Topics Covered

Environmental restoration Robotics Artificial intelligence Ecological data Climate change and agriculture Food systems Biodiversity loss Regenerative agriculture Sustainability

Disclaimer: We provide independent summaries of podcasts and are not affiliated with or endorsed in any way by any podcast or creator. All podcast names and content are the property of their respective owners. The views and opinions expressed within the podcasts belong solely to the original hosts and guests and do not reflect the views or positions of Summapod.

Quick Takeaways

  • Tom Chi argues that the economy is not in opposition to nature but is literally a subset of the ecology, since everything in the economy is ultimately mined or grown.
  • Humanity currently extracts over 90 billion tons of material per year, with per-capita use far higher in Europe and especially the United States than in Asia overall.
  • Most industrial mining, refining, and growing methods are based on decades-old technologies, creating large ecological damage that newer tools like robotics and AI can help overcome.
  • Advanced battery recycling can return materials to better-than-virgin quality at lower cost, enabling a shift from virgin mining to closed-loop material use.
  • Regenerative agriculture and soil diagnostics, combined with AI-guided crop breeding, can improve soil health, reduce inputs, and create crops resilient to climate instability.
  • Satellite-based sensor fusion and long-term biomass data enable precise monitoring of where terrestrial ecosystems are degrading or recovering.
  • Autonomous drones can plant mangroves at extremely high rates, achieving high survival and establishment rates over large areas with very small teams.
  • The ReefGen robot demonstrates that relatively low-cost underwater robots can plant corals and seagrasses at scale, making ocean restoration more accessible and scalable.
  • Designing industrial machines with the explicit intent to repair ecosystems can transform the relationship between technology, the economy, and the natural world.
  • Scaling restoration robotics could allow even a single well-funded actor to deploy thousands of units and have meaningful impact on ocean ecosystems.

Podcast Notes

Podcast introduction and framing of the talk

Host and show identification

Elise Hu introduces the show as TED Talks Daily and notes that it brings new ideas to spark curiosity every day[1:59]
• She identifies herself by name as the host

Framing question and introduction of Tom Chi

Elise poses the question of what if everything we thought we knew about the relationship between the economy and the environment was wrong[2:03]
• She sets up the talk as exploring a paradox where people claim to love nature but collectively destroy it
She introduces Tom Chi as an impact investor who explores this paradox[2:12]
• She summarizes that he will challenge what he views as false trade-offs between economic growth and ecological health and show how innovation can be a force for restoration rather than extraction

Paradox of loving nature while destroying it

Tom describes his focus over the last decade

Tom states he has spent the last decade working on resolving a profound paradox[2:40]
• The paradox is that individuals express love for nature while their collective civilization destroys it at planetary scale

Evidence of individual love for nature

Tom says if you asked 100 random people on the street how they feel about nature, you would get extremely positive answers[3:24]
• He notes people describe nature as inspiring and the most beautiful thing that exists
• He points out that people use nature images as backdrops on their desktops and phones as further evidence of affection for nature
Tom emphasizes that every single person you ask will give a pretty positive response about nature[3:15]

Collective destruction despite positive attitudes

Tom contrasts individual attitudes with collective behavior, saying that as a civilization we are destroying nature at a planetary scale[3:27]
• He identifies this contrast as the core paradox: nature-loving individuals have created an industrial economy that mounts a planetary-level assault on nature

Challenging the economy versus ecology mental model

Origin of the paradox in a flawed mental model

Tom suggests the paradox stems from a broken mental model that has been unconsciously adopted[3:52]
• He expresses the desire to start dissolving that broken model in the talk
He defines the mental model as the belief that economic wins are always traded off against the ecology[4:02]
• In this view, if you want to make the ecology healthier, you must trade off economic gains, and vice versa
• He describes it as a balancing act where attention to either the economy or the ecology swings a pendulum or scale in one direction or the other

Physical reality: economy as subset of ecology

Tom says that while the trade-off framing is a psychological position one can take, it is not particularly physically true[4:31]
• He notes that he is formally trained as a physicist and therefore focuses on what is physically true
He asserts that the economy is not versus ecology; instead, the economy is a subset of the ecology[4:45]
Tom proposes that listeners can quickly prove this to themselves by examining the clothes on their bodies or the objects and media around them[4:57]
• He states that everything the economy has produced is either mined or grown, meaning it comes directly from nature with no abstractions
• Examples he gives include cotton clothing (grown) and metal in a chair (mined)
Tom anticipates the objection that we are moving to a digital or virtual economy and responds to it[5:22]
• He argues that every line of code runs on a physical substrate that was mined or grown
• He notes that every digital service uses server architectures that are mined or grown
• He reiterates that when he says the entire economy is mined or grown, he means it literally and that nothing in the economy fails to come directly from nature
Tom concludes that damaging the ecology inevitably creates problems for the economy[5:47]
• He warns that thinking in terms of a simple balancing act between economy and ecology will cause us to miss the right way to fix these problems

Scale of current extraction and outdated industrial methods

Quantifying global mining and growing

Tom states that at this point in history, humanity is mining and growing over 90 billion tons per year[6:05]
• He translates this to about 11.5 tons per person per year on average
He notes that 11.5 tons per person per year is the average in Asia[6:08]
• He adds that Europe uses about twice that per person, and America about three times that, reacting with a sarcastic "Yay!"
Tom points out that about 2 billion people live on less than $5 a day and use substantially less material, which is why global averages balance out[6:23]
• He reiterates that this describes how much we are mining and growing to feed the economy

Outdated industrial paradigms and opportunity for new tools

Tom reminds listeners that literally everything in the economy is mined or grown[6:42]
• He explains that to power the economy, we have been using very old industrial ideas and processes
He notes that most current growing practices were invented about 50 years ago[6:50]
• He says that most methods of mining and refining metals were invented about 100 to 150 years ago
• He emphasizes that these are not technologies we have updated recently
Tom argues that with new robotic and AI tools, it is the right time to ask whether we could mine and grow differently[7:12]
• He frames the goal as honoring the idea that the economy is a subset of the ecology when redesigning these processes

Tom Chi's background in technology and its relevance

Professional experience in advanced technologies

Tom explains that his career has focused on new inventions in robotics, artificial intelligence, and advanced algorithms[7:29]
• He says he has shipped products ranging from Microsoft Office to web search to self-driving cars
He characterizes these as relatively sophisticated technologies[7:42]
• He suggests his background positions him well to look at mining and growing problems and take a different swing at them
Tom previews that he will share a number of examples that illustrate his approach[7:53]

Three major shifts for aligning economy and ecology

Overview of the three shifts

Tom says the examples he will share fall into three major shifts related to mining and growing[7:58]
• He references three images representing mined materials, closed-loop materials, and ecological repair

Shift 1: More ecological mining and closing material loops

Tom says we need to figure out more ecological ways to mine materials and to get the most out of the ores we extract[8:11]
• The goal is to disturb earth and watersheds as little as possible during mining
He adds that an even better approach than more ecological mining is to avoid mining altogether when possible[8:22]
• To do this, he advocates closing loops through skillful mechanical or chemical recycling so a greater fraction of industrial feedstock comes from recycled rather than virgin materials

Shift 2: Transforming how we grow through regenerative practices

Tom observes that many current growing methods are very unsustainable and damage soil function[8:51]
• He notes that agricultural practices have been wearing down topsoil on lands around the world
He mentions his friend Gabe Brown, from whom he has learned a great deal about regenerative agriculture[9:06]
• From this learning, he concludes that investing in soil function can make growing easier, cheaper, and higher margin every year
• He stresses that regenerative approaches can regenerate soil function, support biodiversity services, and heal the hydrological function of soils

Shift 3: Large-scale ecological repair

Tom argues that after a couple of centuries of the Industrial Revolution, many landscapes have been heavily degraded[9:36]
• He says that if we are serious about renewing the ecology to support a vibrant economy, we will need better tools for large-scale repair

Closed-loop materials: advanced battery recycling

Lithium NMC battery recycling plant example

Tom describes an image from inside the largest lithium NMC battery recycling plant in North America, operated by a company he works with[9:57]
• He presents this plant as a strong example of moving toward a closed-loop world
He explains that the plant uses an advanced form of chemical recycling to process used battery materials[10:06]
• Most lithium batteries have a roughly 10-year life, after which they can no longer serve in cars or consumer electronics and their materials must be recovered
Tom notes that this recycling process is about two times cheaper than the next closest process[10:27]
• He says it can return material to complete virgin quality, describing it as better than what could be mined from the ground in the first place
He highlights that batteries are large, discrete objects that can be collected via reverse logistics, facilitating loop closure[10:46]
• He emphasizes that with robotics and AI for advanced mechanical recycling, or with advanced chemical recycling, we can skillfully increase the fraction of feedstock coming from post-consumer or post-industrial waste streams

Regenerative agriculture and AI-assisted crop development

Mini-renaissance in regenerative practices

Tom says we are at a compelling point in history because there is a mini-renaissance in regenerative agriculture happening globally[11:19]
• He notes that farmers around the world are discovering benefits of agroforestry, intercropping, no-till agriculture, and other practices
• These practices help establish healthy soil function and a healthy soil microbiome
Tom explains that practitioners are now able to measure compelling compounds from the soil[11:43]
• He describes this as allowing the soil to "speak" to farmers, effectively indicating the next actions needed to improve its health
He contrasts this with the past, when soil behavior was more of a black box that had to be interpreted indirectly[11:54]
• Now farmers can have a direct relationship with their soils and manage them skillfully toward greater health, fewer inputs, and higher margins each year

Historical development of crops and new AI tools

Tom recounts the development of corn (maize) as an indigenous project over hundreds of years[12:18]
• He explains that it started from an inedible bit of grass, since corn is a type of grass, and through selective breeding many varieties were created
• Through generations of breeding, people produced high-calorie-per-growth-cycle corn that now feeds a large percentage of global calories
Tom notes that many foods we eat today were selectively bred over time to become large, healthy, and nutritious[12:51]
He introduces a company he has been working with that uses artificial intelligence and machine learning to speed up this selective breeding process[12:58]
• He clarifies that this is not genetic modification; instead, it relies on sequence information from existing commercial crops and native varietals
• The system infers gene functions and maps precise cross-breeding pathways to achieve desired traits

Examples of AI-guided crop traits

Tom lists adaptive sugarcane as an example outcome, which dramatically reduces deforestation required for a given yield[13:30]
He describes heat-resistant tomatoes that can grow in much hotter, drier conditions[13:37]
• He emphasizes this is important because we face at least a 50-year period of destabilizing farmlands as the climate becomes hotter, colder, wetter, and drier in different places
Tom also mentions cotton varieties that are drought tolerant, requiring a fraction of the water-about one tenth-as well as much less pesticide and fertilizer input[14:06]
• He notes that these traits are fantastic for the planet and also for maintaining viable food and material supplies in a destabilized growing environment

Ecological monitoring via satellite and sensor fusion

Company using sensor fusion for biomass mapping

Tom introduces a company called Chloraschia Spatial as an example of scalable restoration-related technology[14:24]
• He says they have done deep work on sensor fusion across multiple satellite feeds
He explains they paired satellite analysis with over a decade of ground-truthing data in jungles, collected meter by meter[14:38]
• This combination allows them to verify how much terrestrial biomass corresponds to remote sensing signals from satellites
Tom says that with this approach they have created the most accurate historical and current assessments of above-ground biomass[14:53]
• Their data show how much terrestrial biomass exists on Earth, with records stretching back to the beginning of the 21st century-over 20 years of data
He notes this allows us to see which landscapes are hurting and which are recovering[15:06]
• For people developing restoration or carbon projects, this provides a way to stay on top of how those efforts are progressing

Limitations of monitoring alone

Tom acknowledges that while this is great technology using advanced algorithms, it is somewhat passive[15:20]
• He points out that monitoring does not by itself restore forests or grasslands; it only tracks change

Reimagining industrial machines as agents of ecological repair

From accidental damage to intentional repair

Tom asks what would happen if we challenged the current linkage between industrial machines and nature[15:43]
• He characterizes the current relationship as an accidental relationship of damage
He proposes imagining an intentional relationship of active repair between machines and ecosystems[15:51]
• He announces he will show two examples that embody this idea

Mangrove restoration with planting drones

Drone planting process and performance

Tom describes a short video where listeners hear ticks, each tick representing a mangrove seed being planted[16:02]
• He states that the pace of planting is about 100 mangroves per minute from a single drone
He reports that two months after planting, over 90% of seeds reach germination[16:11]
• After 14 months, the landscape is fully established with over 85% of the planted mangroves fully establishing

Scalability of drone-based mangrove restoration

Tom emphasizes the scale of the technology, noting that four people can plant over 80 hectares of land in a day using the system[16:34]
• He quantifies this as 120,000 mangroves planted and over 100,000 establishing in one day
He highlights that with robotic scale, human action and intention can be multiplied to rewrite landscapes[16:49]

Track record of the mangrove restoration company

Tom shares that he has worked with this company for about a decade[17:00]
He notes that beyond mangroves, they have restored 20 different terrestrial ecosystems on four continents[17:08]
• Ecosystem types include dry land, inland, mountainous, near-shore, and others
• Their breadth of work inspired Tom to consider applying similar approaches below the water

Underwater restoration with the ReefGen robot

Introduction to ReefGen and its capabilities

Tom introduces ReefGen as a robot line he founded and for which he was the original electrical engineer[17:23]
• He describes ReefGen as its own kind of planting drone designed for underwater environments
He says this robot line is the first in the world to plant live corals back into a coral reef[17:37]
He adds that it is also the first in the world to plant live seagrasses back into seagrass meadows[17:40]
• ReefGen can plant underwater vegetation both as live plants and in seed form
Tom notes that the robot has planted 10,000 seagrass seeds in a single day[17:47]
• He equates this to covering an entire underwater acre with one robot in one day

Cost considerations and scalability of ReefGen

Tom explains that when he discussed building an underwater restoration robot, people advised budgeting about $2 million per robot[18:14]
• He recalls personally thinking more along the lines of $5,000 per robot
He states that the actual cost of the current ReefGen robot is more like $10,000[18:14]
• He points out that this is far less than $2 million and was intentionally kept low to make the technology accessible
Tom emphasizes that communities with nearshore ecosystems needing restoration, whether coral or seagrass, should be able to access such robots[18:38]
• He stresses that affordability is important so the technology is scalable from a capital expenditure perspective
He notes that a single billionaire could spend $50 million and obtain a fleet of 10,000 ReefGen robots[18:46]
• He argues that such a fleet would represent meaningful scale for ocean restoration of many types

Technical details of seagrass planting mechanism

Tom clarifies that the end effector being shown is a seedling-planting end effector, not a seed-planting one[19:04]
• He explains there are two ways to plant seagrasses: from seed and from saplings that grow rhizomically
He describes certain seagrasses that prefer to be planted as saplings and then send out lateral rhizomes, from which new grasses grow upward[19:19]
• The system uses a stake into which seagrass seedlings are placed
Tom explains that the seedlings in the stake feed through a tube into a hopper in the robot[19:30]
• With the current layout, one robot can plant about half an acre of seedlings per day
He says the next version of ReefGen is expected to plant between one acre and an acre and a half per day per robot[19:37]

Closing reflection: using robotics and AI to support ecology and economy

Revisiting the mental model and future vision

Tom returns to the mental model of economy versus ecology and argues for digging into it in a different way[19:56]
• He advocates taking the best tools from the current economy-robotics and AI-and intentionally using them to support ecology
He concludes that by doing so we can build both a healthy planet and a healthy economy for the future[20:08]
• He ends his talk by thanking the audience

Outro and contextual information about the talk

TED event context and production credits

The host notes that the talk was given by Tom Chi at a TED Countdown event in New York in partnership with the Bay Earth Fund in 2025[20:26]
She mentions that TED Talks Daily is part of the TED Audio Collective[20:37]
• She credits the TED Research Team with fact-checking the talk
She lists members of the production and editing team by name and notes that the episode was mixed by a named engineer with additional support from others[20:50]
• Elise Hu says she will be back with another idea and thanks listeners for listening

Lessons Learned

Actionable insights and wisdom you can apply to your business, career, and personal life.

1

The economy is physically a subset of the ecology, so damaging natural systems ultimately undermines economic stability and growth rather than being a separate trade-off decision.

Reflection Questions:

  • • Where in your work or life do you still think in terms of "economy versus environment" instead of recognizing their physical interdependence?
  • • How might your decisions change if you viewed every product, service, or process you rely on as ultimately mined or grown from nature?
  • • What is one concrete decision you could make this month that aligns your economic goals more directly with ecological health?
2

Updating old industrial processes with modern robotics and AI can transform extractive systems into closed-loop and regenerative ones, reducing ecological damage while lowering costs.

Reflection Questions:

  • • Which processes in your organization or industry are still running on decades-old assumptions or technologies that could be rethought with current tools?
  • • How could automation or data-driven insights help you use fewer virgin resources or create less waste in what you do?
  • • What is one legacy process you could map out and intentionally redesign for circularity or regeneration over the next year?
3

Investing in underlying system health-like soil function in agriculture-creates compounding benefits: fewer inputs, higher margins, and more resilience to external shocks.

Reflection Questions:

  • • What is the "soil" in your domain-the foundational system or capability that quietly supports everything else?
  • • How would your results change if you shifted more effort from short-term outputs to strengthening that foundational system?
  • • What specific investment in system health (training, maintenance, infrastructure, relationships) could you commit to over the next quarter?
4

Designing technologies explicitly for large-scale ecological repair can unlock restoration at scales far beyond what manual human effort alone can achieve.

Reflection Questions:

  • • In your field, what kinds of positive impact are currently limited mainly by labor or speed rather than by knowledge?
  • • How could you repurpose existing tools or develop new ones so that each unit of effort produces an outsized, restorative effect?
  • • What small prototype or pilot project could you start that tests using technology to repair, rather than merely exploit, a system you care about?
5

Affordability and scalability are crucial design constraints if you want impactful environmental technologies to be adopted widely by communities that need them most.

Reflection Questions:

  • • Where might high costs or complexity be preventing broader adoption of beneficial solutions in your context?
  • • How could you simplify, modularize, or re-engineer a solution you use so that many more people or organizations could access it?
  • • What is one product, service, or process you influence that you could deliberately redesign to be 10x more accessible over the next few years?

Episode Summary - Notes by Jamie

The army of autonomous robots restoring nature | Tom Chi
0:00 0:00