Anil Madhavapeddy

What I learnt at the National Academy of Sciences US-UK Forum on Biodiversity / Jun 2025 / DOI

I spent a couple of days at the National Academy of Sciences in the USA at the invitation of the Royal Society, who held a forum on "Measuring Biodiversity for Addressing the Global Crisis". It was a packed program for those working in evidence-driven conservation:

Assessing biodiversity is fundamental to understanding the distribution of biodiversity, the changes that are occurring and, crucially, the effectiveness of actions to address the ongoing biodiversity crisis. Such assessments face multiple challenges, not least the great complexity of natural systems, but also a lack of standardized approaches to measurement, a plethora of measurement technologies with their own strengths and weaknesses, and different data needs depending on the purpose for which the information is being gathered.

Other sectors have faced similar challenges, and the forum will look to learn from these precedents with a view to building momentum toward standardized methods for using environmental monitoring technologies, including new technologies, for particular purposes. -- NAS/Royal Society US-UK Scientific Forum on Measuring Biodiversity

I was honoured to talk about our work on using AI to "connect the dots" between disparate data like the academic literature and remote observations at scale. But before that, here's some of the bigger picture stuff I learnt...

Read full note... (2343 words)

We become Junior Rangers at Shenandoah / May 2025 / DOI

What might a Dame of the Realm, a Fellow of the Royal Society, the latest member of the UK Joint Nature Conservation Committee, and me all covet? That's right: a Junior Ranger badge from Shenandoah National Park! After an intense few days, Bill Sutherland, Julia P.G. Jones, EJ Milner-Gulland and I headed into nature to experience the spectacular landscapes of the Blue Ridge Mountains in Virginia and do some birding.

The National Park Service in the US runs a wonderful program for anyone aged 8+ (which we just about qualified for) to introduce people to nature, and Shenandoah is no exception. We visited the local ranger lodge in the park, and picked up a program booklet. They're full of activities for kids to do, but of course adults also pick up a lot of random knowledge (such as the endemic salamander species in the region).

Read full note... (386 words)

Out-of-the-box LLMs are not ready for conservation decision making / May 2025

Our paper on how the careful design of LLMs is crucial for expert-level evidence retrieval has been published today in PLOS One and is available fully open access!

Our findings suggest that, with careful domain-specific design, LLMs could potentially be powerful tools for enabling expert-level use of evidence syntheses and databases. However, general LLMs used "out-of-the-box" are likely to perform poorly and misinform decision-makers. By establishing that LLMs exhibit comparable performance with human synthesis experts on providing restricted responses to queries of evidence syntheses and databases, future work can build on our approach to quantify LLM performance in providing open-ended responses.

In a nutshell, we tested 10 LLMs with six different retrieval strategies on their ability to answer questions related to conservation, benchmarked against the Conservation Evidence database that has been hand-assembled by experts over the last two decades. In some of the retrieval scenarios, models were only allowed to use their pretrained knowledge, whereas in others they had access to the relevant parts of the hand-curated database.

We found that language models had very varying results when relying only on their pretrained data, and were particularly bad at answering questions about reptile conservation. However, given some extra training with the CE database, their performance improved dramatically. When we put these models head to head with human experts (from the conservation evidence team), with a set of questions and with RAG access to the database, we found that the models were just as good as our experts, but answered the questions much much much faster (near instant).

Essentially, LLMs without extra training are likely to perform poorly and misinform decision-makers. This is crucial when considering how to build AI infrastructure for public policymaking.

Read full note... (377 words)

Learnings from the Cambridge Environmental Sustainability Committee / May 2025 / DOI

I joined Cambridge's loftily named Environment Sustainability Strategy Committee this academic year, and have attended a couple of meetings with the latest one being held today. While a lot of what goes on is intricately tied into the University's rather special governance structure and the complexity of the College system, there has been significant progress on making all of this more visible more widely.

Sally Pidgeon, our wonderful head of Enviromental Sustainaibility, has been redeveloping the public website and has put a lot of interesting data online. There is now a new Environmental Sustainability website that tracks the University committment structure more closely, with the areas broken up into Carbon & Energy, Travel & Transport, Waste & Circular Economy, Biodiversity, and Water usage.

Read full note... (842 words)

Humans are the ones that will save nature, helped by AI / May 2025 / DOI

In my earlier note about how AI should unite conservation, I talked about the robust debate ongoing within Cambridge about whether or not we're too "AI obsessed" and are losing track of our goals in the rush to adopt learning algorithms. Jacqueline Garget has written a brilliant roundup about how colleages like Sam Reynolds, Chris Sandbrook and Sadiq Jaffer in the CCI are leading conversations to make sure we advance with eyes wide open.

Read full note... (537 words)

Technology needs to unite conservation, not divide it / Apr 2025 / DOI

I had a tremendous time participating in last year's horizon scan of AI and Conservation, which laid out the opportunities that technological progress from AI (a catchall phrase here) could bring to hard-working conservation practitioners. Since then, there's been a lot of corridor conversations about future projects (and even dinner with the Wildlife Trusts). However, there has also been discussion about the potential harms of our work, most notably in a response letter to our paper written by Katie Murray and colleagues.

Murray et al make two really important points:

• [...] importance of ecological expertise must be recognised as much more than just the expert annotation of training data
• [...] effort should be made to build capacity for AI development in the Global South, so that the rewards of successful research can be shared -- The potential for AI to divide conservation

Read full note... (1481 words)

2nd Programming for the Planet workshop CFP out / Apr 2025 / DOI

Dominic Orchard and I had a blast running the first PROPL workshop a couple of years ago, with a full room and engaged audience in POPL in London. Last year, our sister conference LOCO took over, and it's our turn again this year! PROPL will return for a second outing in October, co-located with ICFP/SPLASH in Singapore in October. Read the call for papers here (deadline 3rd July 2025).

Read full note... (565 words)

New preprint on benchmarking ultra-low power neural accelerators / Mar 2025

Josh Millar just released our latest preprint on how to make sense of the growing number of dedicated, ultra-low-power 'neural network accelerators' that are found in many modern embedded chipsets. My interest in this derives from wanting to decouple from the cloud when it comes to low-latency local environments, and this needs fast tensor operations in hardware. Josh found a huge number of interesting NPUs in modern low-cost chips, ranging from ESP32-based boards over to ARM ones. All of these have quite a variety of tradeoffs, from the operations supported (which affects which models can be run on them) to the amount of memory and CPU power. This is the first comparative evaluation and independent benchmarking of several commercially-available micro-NPUs. We developed an open-source model compilation framework to enable consistent benchmarking across diverse hardware, measuring end-to-end performance including latency, power consumption, and memory overhead. The analysis uncovered surprising disparities between hardware specifications and actual performance, including unexpected scaling behaviors with increasing model complexity.

LIFE becomes an Official Statistic of the UK government / Mar 2025 / DOI

Our recently published LIFE biodiversity metric has just been integrated into a newly recognised Official Statistic from the UK government! This integrates the core LIFE biodiversity metric with food provenance data to track the environmental impacts of our consumption habits.

Read full note... (629 words)

A fully AI-generated paper just passed peer review; notes from our evidence synthesis workshop / Mar 2025 / DOI

Access to reliable and timely scientific evidence is utterly vital for the practise of responsible policymaking, especially with all the turmoil in the world these days. At the same time, the evidence base on which use to make these decisions is rapidly morphing under our feet; the first entirely AI-generated paper passed peer review at an ICLR workshop today. We held a workshop on this topic of AI and evidence synthesis at Pembroke College last week, to understand both the opportunities for the use of AI here, the strengths and limitations of current tools, areas of progress and also just to chat with policymakers from DSIT and thinktanks about how to approach this rapidly moving area.

Read full note... (1565 words)

A trio of papers I read on biodiversity and forests this week / Feb 2025 / DOI

This week I've been reading three really nice pieces of work by my colleagues, in the form of a review paper on biodiversity and AI, a benchmark for 3D forest reconstruction using laser scanners and a mobile app for measuring the width of tree trunks. A real bonanza for forest lovers!

Read full note... (793 words)

Updated preprint on quantifying biodiversity cost of food consumption / Feb 2025

We've uploaded a revised preprint on our ongoing work on quantifying the biodiversity cost of global food consumption, lead by Thomas Ball. This is based on the recently published LIFE metric, combined with supply chain data and provenance modeling.

Read full note... (196 words)

Disentangling carbon credits and offsets with contributions / Feb 2025 / DOI

The terms carbon credits and carbon offsets are often used interchangeably, but are in fact two distinct concepts. I've spent a nice Sunday morning reading up on some recent articles that Bhaskar Vira sent me which introduce a third term, known as "carbon contributions". Rather than this adding confusion, I found it helped me clarify my own thoughts on the matter, which I note down here in draft form. (Update 7th Feb: I've revised this several times after many discussions this week, especially with David Coomes and Srinivasan Keshav, with full list of credits in the end)

Read full note... (3552 words)

Position paper on scientifically credible carbon credits (via 4C) / Jan 2025 / DOI

My colleagues Thomas Swinfield and Eleanor Toye Scott lead the publication of a comprehensive report of the steps the voluntary carbon market needs to take to restore its scientific credibility, with input from many of us in 4C and beyond.

• establishing common standards for carbon quantification and accounting, to cover additionality, leakage and permanence.
• avoiding perverse incentives and align the motivations of all stakeholders with high-integrity outcomes. [...]
• issuing all carbon credits based on trusted primary observations.
• making all the data needed to reproduce carbon calculations available in standard file formats.
• [...] reporting social and biodiversity dimensions of projects separately from carbon calculations.
• integrating DMRV methods into carbon and biodiversity accounting standards to reduce the financial and administrative burdens on nature-based projects and the local communities participating in or affected by them.

Read full note... (186 words)

Updated preprint on LLMs for evidence-based decision support / Jan 2025

We have just updated our preprint on using LLMs for evidence decision support with more evaluation results and corrections from peer review.

Our findings suggest that, with careful domain-specific design, LLMs could potentially be powerful tools for enabling expert-level use of evidence syntheses and databases. However, general LLMs used "out-of-the-box" are likely to perform poorly and misinform decision-makers. By establishing that LLMs exhibit comparable performance with human synthesis experts on providing restricted responses to queries of evidence syntheses and databases, future work can build on our approach to quantify LLM performance in providing open-ended responses.

See also the fantastic EEG seminar talk that the student group who worked on this over the summer gave towards the end of last year.

LIFE metric published in Royal Society Phil Trans B / Jan 2025

After some years of hard work, our Mapping LIFE on Earth biodiversity metric was published today in a special issue of the Royal Society Philosophical Transactions B! The idea behind LIFE is that although human-driven habitat loss is known to be the greatest cause of the biodiversity crisis, we do not yet have robust spatially explicit metrics that quantify the relative impacts of human actions on species extinctions. And that's what LIFE provides: a way to compare the relative impacts of some landuse anywhere in the world, in a manner that is globally applicable.

Read full note... (409 words)

Published a legal perspective on high integrity forest carbon credits / Nov 2024

Sophie Chapman lead an effort to explore a novel legal framework for forest carbon credits that separates carbon tenure (i.e. title and associated property rights to the land and trees which store the carbon) from the carbon rights (i.e. title and associated rights to monetise and manage the credits which symbolically represent the carbon stored in the trees), while also specifying the relationship between the carbon tenure and the carbon rights.

The resulting paper has just been published in the Climate and Carbon Law Review journal, and is available as open access for your perusal.

Preprint on using LLMs to for evidence-based decision support / Nov 2024

We have just uploaded a preprint on using LLMs for conservation evidence, based on our work on large-scale crawling of the academic literature. Well done in particular to Radhika Iyer for having done the bulk of the evaluation on this as part of a very productive summer internship with us! This work evaluates whether LLMs can facilitate evidence-based decision support for conservation by testing ten different LLMs against human experts on multiple choice questions about conservation interventions. We found that with careful design, open-book LLM performance was competitive with human experts on filtered questions, both in correctly answering them and retrieving the source documents. However, general LLMs used "out-of-the-box" performed poorly, highlighting the importance of domain-specific design to avoid misinforming decision-makers.

Mapping greener futures with planetary computing / Oct 2024

I got invited by Sertaç Sehlikoglu to deliver a lecture to the Masters students down at the UCL Institute for Global Prosperity. I talked about the recent work on planetary computing, with an overview of the LIFE and FOOD papers.

Royal Society meeting on ecological/commercial risks / Oct 2024 / DOI

I'm at the Royal Society this morning for the 2 day programme on "How does ecological risk related to commercial risk?", and am reporting on the morning session. The full program is being livestreamed so please do dial in if the below notes seem interesting to you. I put this note up almost live, so any errors below are my own. (Update: partial day 2 notes now available below)

Read full note... (6736 words)

Mitigating credit reversal risks in nature-based solutions / Sep 2024

Many of the questions around our recent Nature Sustainability commentary on NbS credits revolve around how to finance new projects if credible credits need to be ex-post. Our latest paper published in Carbon Management on "Mitigating risk of credit reversal in nature-based climate solutions by optimally anticipating carbon release" tries to address this.

The problem with selling ex-ante (future) carbon credits for (e.g.) a deforestation avoidance scheme is that project reversals can happen in the future ("deforestation has increased") thus rendering any credits issued previously useless. On the flip side though, an overly conservative view of the future ("the entire forest will disappear overnight!") is clearly so conservative that it doesn't serve the best interests of the project developer. So ideally, a project would make realistic but conservative ex-ante predictions that is safe for both project developer (who gets more funds upfront) and a carbon credit purchasers (who needs to account for impermanence of nature credits).

Our paper shows how to do this by calculating a "release schedule" to predict future drawdowns, and then issuing extra credits when the release at some future date is less than predicted by the release schedule. We use verified ex-post observations to construct these release schedules, and design them to bound the risk of the project becoming negative overall (that is, net drawdown is negative) and thus failing.

Read full note... (355 words)

Building species models of the planet / Sep 2024

I don't normally announce funded grants (preferring to focus on outcomes), but I'm really excited by this one and couldn't resist! Myself and my colleagues Srinivasan Keshav (from computer science), David Coomes (from Plant Sciences), Andrew Balmford (from Zoology) and Neil Burgess (the Head of Science at UNEP-WCMC) have just received a £1.2m grant from the UKRI to work on building foundation models for planetary intelligence.

Now, normally a grant isn't news, but I wanted to highlight the scheme that it came under. UKRI announced an interdisciplinary program specifically for projects that don't normally get funded by just one research council. In our case, this work usually falls between the cracks of EPSRC ("too much nature") or NERC ("too much engineering") or STFC ("not enough satellites"). But this interdisciplinary program expressly assembled a panel across all these areas, and collectively gave us a shot. I really hope this scheme continues to gather steam within the UKRI.

As to what we're doing? There'll be the evolution of the work described in Remote Sensing of Nature and Mapping LIFE on Earth, with lots of domain knowledge that we're pulling together with our partners at UNEP-WCMC (especially Neil Burgess and Ian Ondo) on plant and animal species distributions across the globe.

Read full note... (339 words)

Nature Sustainability commentary on carbon and biodiversity credits / Aug 2024

Our commentary on nature-based credits has been published in Nature Sustainability, lead expertly by my colleagues Thomas Swinfield and Sophus zu Ermgassen.

In our view the carbon credits markets are vitally important for forest conservation, but the key is to only transact these credits after they have been proven to be demonstrably additional using robust statistical techniques, so that we know before a sale that each credit represents real gains that would not otherwise have occurred without the carbon finance.

A more scientific approach that supports transparent, third-party validation could absolutely transform these markets. And given the rapid rate of tropical forest loss, such upscaling of credibility is vitally necessary to raise investor confidence in protecting nature, since we can now be confident that every "credit" sold is resulting in real climate benefit. There are real questions remaining about this reform, of course.

Read full note... (509 words)

COMPASS 2024 report on the CoRE stack RIC meeting / Jul 2024 / DOI

This is a trip report of ACM COMPASS 2024 held in New Delhi, which had a novel track of "Research to Impact Collaboratives" that drew me in. The general chair, Aadi Seth wrote a fantastic book on "Technology and Disempowerment" a few years ago, and he organised one RIC session on the CoRE Stack -- a climate adaptation stack for rural communities. This was a must-visit for me as it is closely related to the work we've been doing on Remote Sensing of Nature and Planetary Computing. The following notes are somewhat raw as they have only been lightly edited, but please refer to the more polished documents on the agenda for ACM COMPASS RIC and the overall CoRE Stack initiative on commoning technologies for resilience and equality

The conference itself was held at IIIT-D in New Delhi, right at the cusp of the monsoon season and after record-breaking temperatures. Luckily, as always, the hospitality and welcoming nature of New Delhi overrode all the climate discomfort!

Read full note... (4361 words)

Second preprint of the LIFE biodiversity metric available / Jul 2024

We have made an update to the LIFE biodiversity metric based on reviewer feedback, and are very pleased that it has been accepted for publication early next year as part of a special issue from the Royal Society. Any comments would be most welcome before we submit the final proofs in a few months. The revisions incorporated feedback from peer reviewers and represent an improved version of the methodology. Getting accepted for the Royal Society special issue is particularly exciting as it means the LIFE metric will be published alongside other important contributions to biodiversity science. This metric has already proven useful in our work on food systems and will hopefully become a widely-used tool for quantifying biodiversity impacts across diverse applications.