The Bletchley Declaration – The First AI Safety Summit
It is argued that the biggest issue of potential opportunities and risks that face humanity is the introduction of “frontier AI” and particularly the governance of its future development which will be many times more powerful than it is today.
November 1st & 2nd saw the first AI Safety Summit, hosted by Rishi Sunak and the United Kingdom government, the conference was held at Bletchley Park, the site made famous during World War 2 where codebreakers including Alan Turing the father of the modern computer broke the code of the Enigma machine thus shortening the war by two years. The summit ended with the publication of a communique (The Bletchley Declaration) named after the location of the conference
The conference was attended by delegates from some 27 worldwide governments, including senior diplomats from the U.S. and China along with some of the biggest names from the tech industry and CEOs of top artificial intelligence companies.
There was no attempt to reach an agreement on enforceable guidelines that would govern AI development. However, the Prime Minister did announce the UK government would receive early access to the models to implement safety assessments. He also announced that Yoshua Bengio, Scientific Director of the Montreal Institute for Learning Algorithms, will chair a group that will report on the scientific risks and capabilities of “frontier AI” systems.
The Prime Minister also announced that the U.K.’s AI Taskforce set up a few months ago will become permanent and renamed the U.K. AI Safety Institute with the job of carrying out safety evaluations.
There is little doubt that AI will make some jobs obsolete, but estimations vary widely on what the overall effect will be in the years to come. One of the most interesting debates to emerge from the Summit is the