Sources: the UK plans to announce an international advisory group on AI, loosely modeled on the UN's IPCC, at the November 2023 AI summit in Bletchley Park
The British government wants to draw on expert knowledge about the fast-developing technology — The UK is planning to announce …
An IPCC-style body would add a standing expert-evidence layer to the UK’s summit strategy, following its earlier plan to convene “like-minded” countries on AI. The significance is institutional: it could give recurring assessments a role alongside ad hoc political declarations.
First-order effects
The UK would gain a proposed mechanism for collecting and presenting international expert advice on AI, tied to its summit agenda.
Governments, academics and participating AI companies would face a new forum through which technical risk assessments could be surfaced and compared.
Second-order effects
A standing advisory process could shape the issues future AI summits prioritize, moving discussion from broad principles toward shared evidence and assessments.
Other governments and AI developers may need to engage with the group’s work if it becomes a focal point for international safety discussions, though its authority would depend on participation and mandate.
Third-order effects
The move points toward AI governance built around recurring scientific assessment bodies as well as national rules and summit diplomacy.
If countries sustain such institutions, competition over AI policy could increasingly include influence over how risks are defined, measured and communicated internationally.
The trend: AI governance is shifting from one-off leader summits toward durable institutions that can translate fast-moving technical evidence into shared policy agendas.
NEW: The UK is planning to announce an international advisory group on artificial intelligence at its AI summit next month, loosely modelled on the UN Intergovernmental Panel on Climate Change. W/ @madhumita29 https://www.ft.com/...
The UK's AI Safety Summit is set to highlight the risk that ‘frontier’ models could escape human control or be used to produce catastrophic effects in domains like cybersecurity & biotechnology, according to a draft communiqué seen by @EURACTIV. https://www.euractiv.com/...
Britain will seek a global agreement on the risks of runaway frontier AI at its summit next month. Full draft communiqué and details for POLITICO Pro subscribers: https://pro.politico.eu/...
“Advisory group will comprise a rotating cast of academics and geographical expert professionals who would likely write an annual report on AI developments” Would be good to see it coordinate with the UN's new AI Advisory Body set to determine a model for a new UN AI agency/body
Good idea, particularly as interest in evaluations and benchmarking is growing. One thing the AI world would benefit from is a more empirical approach to risk, and an independent, well-staffed and well funded body can help in this respect. https://www.ft.com/... [image]
Sorry, but we don't need international governance institutions to understand the risks of AI, or to delay national policy solutions. The last 20 years of academic research, advocacy, and media reporting tells us more than enough about the risks of AI. https://www.ft.com/...
V good to see opinion moving in the direction I and others advocated earlier in the year, with an ‘IPCC for AI’ a possible outcome of the summit in a few days https://www.ft.com/... https://www.noemamag.com/...
The only sentence you actually need to read to understand this op-ed: “Before we charge head first into *over-regulating* (...)” https://www.ft.com/...
We need an independent, expert-led body empowered to objectively inform governments about the current state of #AI capabilities by @mustafasuleyman & @ericschmidt https://www.ft.com/...
We need an AI equivalent of the IPCC “This proposal has been developed jointly by Mustafa Suleyman, Eric Schmidt, Dario Amodei, Ian Bremmer, Tino Cuéllar, Reid Hoffman, Jason Matheny and Philip Zelikow” Not. One. Woman. https://www.ft.com/... via @ft
“Regulating research and development in AI is incredibly counterproductive,” LeCun, one of the world's leading AI researchers, said. “They want regulatory capture under the guise of AI safety.” Meta's Yann LeCun argues against premature AI regulation https://www.ft.com/...
Yann LeCun is so frustrating. He must know that this isn't the argument! He's an incredibly clever man! Has no one ever sat him down and explained the basics of why people worry about AI existential risk? Why does he keep attacking this bizarre strawman? https://www.ft.com/... [i…
“Regulating research and development in AI is counterproductive,” @ylecun , one of the world's leading AI researchers, told @FT ahead of next month's conference on AI safety. “They want regulatory capture under the guise of AI safety.” https://www.ft.com/...
“There's no question that we'll have machines assisting us that are smarter than us. And the question is: is that scary or is that exciting?” Meta's Yann LeCun https://www.ft.com/... via @ft
Some interesting comments from Meta's chief AI scientist in an upcoming ep of Tech Tonic. @ylecun says we shouldn't regulate AI research and that fears of existential AI risk are unfounded because intelligence doesn't naturally lead to power-seeking. https://www.ft.com/... [image…