Last week, WWF announced “a new tool to help financial supervisors, asset managers and financial institutions evaluate the net-zero transition plans of the companies they are investing in.” Normally, I wouldn’t give this type of announcement a whole lot of thought, but this one gives me much pause because of two letters: AI.
“The Artificial Intelligence (AI) tool that will automate and scale up the analysis of transition plans based on a proposed methodology is currently under development and will be publicly available and follow the structure of the ChatReport tool.”
Details of the tool are in this report that WWF commissioned from the University of Zurich and the Oxford University’s Oxford Sustainable Finance Group – the first 18 pages of which explain the problem as they see it, and it isn’t until page 26 that the specific algorithm and AI tool is mentioned. To be fair, it seems generally aligned with current or proposed climate disclosure frameworks:
“… we analysed 28 different frameworks, published in the years 2021 (5 frameworks), 2022 (12 frameworks) and 2023 (11 frameworks). We identified more than 250 individual indicators, which we then combined into a condensed indicator framework…”
I don’t know if “condensed” is totally appropriate – the 5-page list of indicators, requirements, priorities, and red flag triggers (pages 21-26) is certainly ambitious. The introduction of the academic paper on the ChatReport tool itself explicitly acknowledges my main concern with AI at this point:
“While AI-powered tools can automatically analyze the data, they are prone to inaccuracies as they lack domain-specific expertise.”
Yet the authors contend they fix this problem by “collaborating with experts in climate science, finance, economic policy, and computer science, demonstrating how domain experts can be involved in developing AI tools.” Maybe, but I’m unconvinced and skeptical. At the very least, I wouldn’t want to be one of the guinea pigs – there is too much at stake if things go wrong. Should any company decide to participate in an effort to help develop the tool, it could end up being quite a rabbit hole. I’m concerned about liberties taken/assumptions made by the algorithm in terms of original data validity and governance of the technology/methodology. Or perhaps I am just a stubborn Luddite who is afraid of Skynet and The Terminator.
If you aren’t already, subscribe to our complimentary ESG blog here: https://practicalesg.com/subscribe/ for daily updates delivered right to you.