Artificial Intelligence (AI) poses a potentially catastrophic threat to humanity, according to the Bletchley Declaration signed by 28 countries, including the United States, EU, China, and the UK at the outset of a two-day technology conference.
The UK convened governments, the academic community, and AI-focused companies to discuss ways to mitigate the technology’s risks and whether it’s feasible to do so.
The meeting was orchestrated by Prime Minister Rishi Sunak, aiming for the UK to mediate between the economic blocs of the United States, China, and the European Union.
Among the 100 attendees of the conference were global leaders, tech titans like Elon Musk, ChatGPT’s chief Sam Altman, and numerous academics.
“There exists potential for severe, even catastrophic harm, whether intentional or unintentional, arising from the profound capabilities of AI models,” the Bletchley Declaration states, released on the inaugural day of the AI summit at Bletchley Park in central England.
Bletchley Park was chosen for the summit, a historic site where British scientists decoded German messages during World War II.
“The Declaration fulfils key summit objectives in establishing mutual agreement and accountability on the risks, opportunities, and the continuation of international collaboration on AI safety and research, especially through enhanced scientific cooperation,” the UK government highlighted in a separate statement accompanying the declaration.
The adopted declaration calls for transparency and accountability from entities developing cutting-edge AI technology.
The Declaration lays out a dual-focused agenda, centring on recognising risks of mutual concern, building scientific understanding of them, and collaborating on policies to mitigate those risks.
“This involves, alongside increased transparency from private entities developing AI capabilities, appropriate assessment metrics, safety testing tools, and the development of pertinent public sector and scientific research capabilities,” the declaration concludes.
Foto: Chris J. Ratcliffe/Bloomberg