#74: Proving carbon removal, anthropomorphising AI, and twinning tech and non-profits
A Q&A with creative technologist Neef Rehman
It’s my turn to write this week, but I couldn’t resist introducing you to another incredible guest.
Meet Neef Rehman.
Neef is a Creative Technologist working in climate. After starting his career in the sciences, he honed his skills in software engineering and design at the renowned digital product studio ustwo. Earlier this year, Neef joined Isometric, a startup creating a verification standard and registry for carbon removal. Alongside his work at Isometric, Neef is a technical advisor to China Dialogue—an independent, non-profit organisation that reports on the world's environmental challenges.
Beyond his career background, I wanted to interview Neef for his way of thinking. Neef is interested in sociology and often explores how technology relates to the underlying facets of human nature. His thoughtful, interdisciplinary approach shines through in our conversation. I hope it sparks as many questions and ideas for you as it did for me.
You describe yourself as a Creative Technologist. How is this role different from, say, a Software Engineer?
This is a great question, and still not one I have a perfect answer to. I see the Creative Technologist role as an intentionally broad space for anyone with technical skills who incorporates a creative or experimental practice into their work. It can refer to someone who's both a designer and developer, or a computer graphics specialist, and facets of the role could include things like applied AI or hardware engineering. It's a little fuzzy, but I've found it a useful title for myself, as I love to work in interaction design, data visualisation, and animation alongside the more typical software engineering tasks. I think the title helps get this breadth across to others.
You recently left a longstanding role at an established digital product studio to join the 2022-founded carbon removal startup, Isometric. What's the most important problem to solve in carbon removal? Why did you decide to join the mission?
I'll answer the second question first, as it's a little easier! I'd been laser-focused on shifting my career to something in climate for a while. It’s something I’ve always cared a lot about, and in recent years has been intersecting more with the tech world. Carbon removal piqued my interest because it presents a range of interesting challenges, which leads me to your first question.
A quick note on Carbon Dioxide Removal (CDR): it's the process of removing CO2 from the atmosphere and locking it away for decades, centuries, and even millennia. There are various methods, from using natural processes like in Enhanced Weathering, to developing new technologies such as Direct Air Capture. These methods aim to reduce the concentration of greenhouse gases in the atmosphere to mitigate the impacts of climate change.
Carbon removal is different from more traditional tree-planting offsets. While those offsets are essentially a promise to remove carbon in the future (as a tree grows over time), carbon removal is about removing carbon that's in the air right now.
Onto what I think is the industry's biggest problem: proof. How do organisations verify that they've removed carbon from the air? If you’re providing something more concrete than a promise, there’s both a much greater capability, and a greater responsibility, to prove it. The field needs to ensure that each claim is watertight, independently checked, and uses methods agreed upon by the academic community. It’s no mean feat.
Proving climate impact is also increasingly in the public eye this year. Reports of "worthless" offsets—via these more traditional tree-planting methods—have come in thick and fast. Rightly, there’s an increasing amount of scrutiny towards all forms of carbon credits as a result. All of this is why we’re building a verification standard and registry at Isometric, to improve trust for the carbon removal industry, scientists, and the public.
Can you give an example of a specific design or development challenge that’s arisen from trying to prove climate impact in this way?
Yes! Providing everyone with the ‘right amount’ of data is a huge challenge. As a public registry, Isometric gets anyone from interested citizens to highly renowned scientists interacting with our platform. The threshold for how much people need to see to build trust can vary wildly as a result. We want to ensure that someone totally new to climate science and someone who is an expert can meet their threshold quickly without being put off by complex maths.
Design-wise, our solution has been to handle data in an incremental way. We start with a high-level description and interactive diagram for each removal method, allowing people to see an overview of the calculation behind each removal. People can then drill down into each facet of that calculation if they like—all the way down to the source document for each data point. This solution comes with plenty of engineering challenges. Take the calculation facets. They can be nested deep within the total removal calculation, for example. Any fans of recursion will know where that’s going.
Anthropomorphing AI distracts us from the intentions of the people behind the systems. When these people are making tools that can radically change the landscape of power, we shouldn’t be taking our eye off the ball.
I want to touch on AI briefly since you recently gave a fascinating keynote where you spoke about AI anthropomorphisation—when we attribute human-like qualities, characteristics, or behaviours to artificial intelligence systems. Why is this an important topic?
I’ve seen a lot of AI discourse over the last year be centred around a perceived, subjective experience that AIs may have.
Are they sentient?
Why do they make specific decisions?
How do I trick them into wanting to do something?
From the sensational reporting around claims of a sentient Large Language Model (LLM) to the conversations about models manipulating people after a chatbot didn't know the current year, we tend to anthropomorphise AI.
In my talk, I wanted to highlight that this is an unhelpful framing. Most researchers, and even LLMs themselves, would agree that this subjective experience doesn't exist. So why should we waste so much time discussing it? To me, anthropomorphising AI is not only unhelpful, but it also leads to problematic outcomes. Ascribing subjectivity to LLMs undermines the agency of those who create them.
Instead of anthropomorphising AI, we should be scrutinising the flow of capital resulting from its deployment, the data sources that make up a model's training, and how these models are then used. Fei-Fei Li, co-director of Stanford's Institute for Human-Centered AI, sums it up nicely in her recent interview with The Guardian:
"To be clear, AI is "promising" nothing. It is people who are promising – or not promising. AI is a piece of software. It is made by people, deployed by people and governed by people."
Anthropomorphing AI distracts us from the intentions of the people behind the systems. When those people are making tools that can radically change the landscape of power, we shouldn’t be taking our eye off the ball.
Now for my ritual questions! What, if at all, do you wish the tech sector understood better about the non-profit sector and vice versa?
Less of a wish and more of an observation, I've recently learned how important the different funding approaches are to how the work gets done. Non-profit funding tends to be project-based; you're expected to see a pre-defined idea through to completion. Tech funding is often product-based; you receive a cheque and evolve the idea through continuous experimentation. One isn't necessarily better than the other, but it's worth remembering that our sectors probably take these respective approaches for granted. It certainly makes for a bigger difference than I initially imagined.
New to Pass It On? Why not join us for free
How can the tech and non-profit sectors come closer together?
I'm not sure I have a great answer to this, but—in the spirit of experimentation—here are a few things I'd be interested to try:
Invite each other to speak at your offices. In my experience, startups love hosting external speakers, but almost always bring guests from other tech companies. Why not invite someone from a non-profit instead, and vice versa?
Offer secondments to non-profits and accept them in reverse. Many companies have volunteering arrangements to provide 'on the ground' time to non-profits. Why not take it one step further and allow people to embed themselves operationally and see what they could learn from non-profit employees in kind?
Create 'twin' relationships between relevant non-profit and tech companies. It's a play on the twinned city concept between British and European towns that aimed to signal cooperation and build exchanges between the countries. I imagine a twinned tech and non-profit sharing training, tools, and people. Why not integrate meaningfully? 'Deeper' is better than 'more' when it comes to things like this, in my opinion.
Which three books or other media have impacted you most and why?
Duty Free Art by Hito Steyerl. I'm usually not one for re-reading books, but I've come back to chapters of this one several times. It's an amazingly broad read that ties together topics like global conflict, fake media, and linguistics into a salient—and highly contemporary—take on the role of art.
Less is More: How Degrowth Will Save the World by Jason Hickel. A thorough history of capitalism, power, and our relationship with nature, combined with an analysis of the economy's impact on the environment. A short read, but one that has made me think a lot about how I consume and the system of capital that we all live in.
Embrace of the Serpent by Ciro Guerra. This is one of my favourite films. Set in the Colombian Amazon, across two time periods, it covers the relationship between an indigenous shaman and a Western colonialist. Each frame of this film could be an award-winning photograph, and it plays with themes of symbiosis and extraction from nature with a deft hand.
Thanks so much for reading,