OpenAI, Google, Microsoft or Meta are not the only ones to have launched a race to play the leading roles in artificial intelligence (AI). Political leaders are also competing for initiatives to supervise – and encourage – this software capable of accomplishing human tasks, such as the text or image generators ChatGPT or Midjourney. Tuesday October 24, the European Union hopes to find a political agreement on the Artificial Intelligence (AI) Act or, at least, compromises on the essentials of this draft regulation.
Then, the 1er and November 2, British Prime Minister Rishi Sunak will welcome representatives of foreign states and tech giants to London for a “international summit on AI safety”. Later in November, G7 countries will meet for a meeting of the “Hiroshima process,” a discussion on AI launched in Japan in May.
This convergence illustrates a feeling of political urgency to seize a technology considered both very promising and worrying. But in this frenzy – the United States, the OECD and China are also active – the strategies differ.
Divergences between European and British approaches
In 2021, Brussels launched the first major legislative project on AI in the world: the AI Act prohibits certain uses (systems of “social rating”THE “subliminal techniques” handling, etc.) and, for uses deemed “high risk” (autonomous driving, sorting of CVs, allocation of bank credits, etc.), imposes obligations, such as minimizing the error rate and discriminatory bias, checking the quality of training data, etc.
On the controversial point of AI models “general purpose” (such as those generating texts or images), the avenue for compromise between the European Parliament and the Member States consists of imposing obligations on the most important models (beyond a computer calculation threshold used for training, or a number of users or business customers in the EU, according to a document cited by Context). Software manufacturers should also ensure that they have taken measures to comply with the “copyright” content used for training.
London, which wishes to become an AI capital, has chosen to focus on the risks considered the most existential: these are linked to a “intentional harmful use” – to generate computer attacks or biological weapons –, or to a “AI control problem” who could escape humans, we read in a draft of the joint summit declaration, mentioned by Euractiv. The statement recalls the alarmist letters calling for ” pause ” AI or deeming it as dangerous as “pandemics or nuclear war”.
You have 50.5% of this article left to read. The rest is reserved for subscribers.