The Global Summit on AI Governance, which will take place on 1er and November 2, 2023, is already raising questions about its effectiveness. It should be based on a voluntary global register of major models.
The United Kingdom is preparing to host a global summit on the governance of artificial intelligence, under the aegis of the local Prime Minister, Rishi Sunak. But a major concern arises from the event: the meeting is focused on apocalyptic scenarios involving uncontrollable AI. For experts, who deplore the “non-committal” aspect of the summit, the cataclysmic approach is not the right one and it would be better to concentrate on short-term problems. You will see that there is no shortage of fears of a predicted flop.
A summit that will depend on the goodwill of the global companies behind the big AI models
The British government, despite its reversals on issues such as climate policy or commitment to oil drilling, communicates a lot about this initiative aimed at raising awareness among the international community of the dangers linked to artificial superintelligence which would be out of control. Scenarios include the production of biological weapons, increased authoritarianism, the threat to democracy and the global financial system.
The summit, planned for 1er and next November 2, is particularly important for Rishi Sunak. The British Prime Minister aims to make the UK the world leader in AI safety regulation. Just that. Except that serious criticism comes to tarnish an event which has not even started yet.
According to Wired, two sources very close to the summit confided that the latter would only consist of creating a global registry based on large voluntary AI models. Therefore, we can doubt the effectiveness of such a meeting, to the extent that it will depend on the goodwill of the large American and Chinese technology companies behind these major models. These include Amazon, Google and Microsoft.
The British feel under-represented at home and do not want to follow European Union rules
The summit also brings to the surface the diplomatic tensions of the moment. For example, the American government does not approve of the arrival of Chinese officials. British AI experts regret that the focus is on potential disasters, relegating to the background the immediate risks and the benefits, because there are also them, of the technology.
In more detail, the summit plans to examine two types of AI: those with narrow but potentially dangerous capabilities; and cutting-edge artificial intelligence, defined by the British as “powerful and versatile”. Many UK AI players feel effectively excluded, as the majority of companies capable of developing such models are Chinese or American. The only player from the United Kingdom is none other than Google DeepMind, but the company founded in London was bought by Google in 2014, to become its AI branch. Does it count or not in your opinion?
Finally, experts regret that the debates on real national regulation of AI are at a standstill, while the European Union is making progress in this area. In a logic of confinement, specialists also believe that the United Kingdom should propose its own rules, instead of following the EU initiative. We have no doubt about the good atmosphere on site in a few days.
Source : Wired
11