2023-11-03 16:53:00
The week which is ending should reassure those who believe that politicians are not sufficiently interested in the progress of artificial intelligence (AI).
It all started on Monday, when Joe Biden signed an executive order aimed at regulating this technology, a first in American history.
Joe Biden signs executive order against malicious uses of AI
Under it, large companies in new technologies resulting in cutting-edge AI will have to notify the US government of their work and share with it the tests aimed at ensuring that this AI is safe. American cloud giants will have to notify the government within 90 days if a foreign actor uses their service to train large linguistic models.
The decree also aims to fight against false content generated by AI, in particular the famous deepfakes, these devilishly realistic fake videos that can put words into a person’s mouth that they have never spoken. “ THE deepfakes use AI-generated audio and video content to build reputations, spread fake news, and commit scams “, declared the American president when signing the decree. This false content generated using AI constitutes a source of even greater concern in the United States as the country is now in the middle of the presidential campaign for the 2024 election.
The Commerce Department will be responsible for helping companies create labels to determine whether or not content was generated by AI, labels which will also be used by different branches of government. Finally, the decree provides for the creation of an AI Safety Center within the Department of Commerce. Responsible for establishing standards for AI security, it will work in cooperation with a similar body to be created in the United Kingdom.
The limits of Joe Biden’s initiative lie in the fact that it is a presidential decree: in order to be able to regulate private practices more deeply, the president will have to have a law passed by Congress, a challenge in a highly charged political landscape and in the midst of an electoral campaign, although AI is one of the areas where Republicans and Democrats can reach agreement.
Meeting at Bletchley Park
For the second big news of the week around AI regulation, head to the other side of the Atlantic, to Bletchley Park, north of London, where the AI Safety Summit took place on Tuesday and Wednesday. A place chosen by Prime Minister Rishi Sunak for its symbolic significance. It was there that, during the Second World War, the mathematician Alan Turing, laying the foundations of modern computing, designed the computer which made it possible to decode Enigma, used by the Germans to encrypt their communications.
The event featured a star-studded guest list, including US Vice President Kamala Harris, Commerce Secretary Gina Raimondo, European Commission President Ursula von der Leyen, and Chinese Vice Minister Wu Zhaohui. of Science and Technology. King Charles, for his part, made a video appearance at the event, while Emmanuel Macron and Olaf Scholz, the German Chancellor, were conspicuously absent. On the private side, there was Mustafa Suleyman, co-founder of DeepMind, as well as star entrepreneur Elon Musk, known for his apocalyptic predictions about AI, who spoke with Rishi Sunak during an interview broadcast live on X at the end of the event.
The British Prime Minister achieved a diplomatic success with the signing of an international declaration recognizing the need to protect against the risks posed by AI, ratified by the European Union as well as 25 countries, including China and the United States. United. The signatories also commit to setting up an AI audit process for companies, similar to what Joe Biden’s decree proposes.
Will Generative Artificial Intelligence eliminate work?
If we stay in the declarative order, this is the first time that China has agreed on the subject with Western governments.
« Seeing Deputy Minister Whu on the same stage as Gina Raimondo sends a very positive signal that governments have understood the need to work together on such an issue. Especially since China adopted regulations on AI this summer from which we can draw inspiration “, says Russ Shaw, founder of London Tech Advocates, a local network of new technology players.
Two future AI Safety Summits are to take place in South Korea and France next year, another success for Rishi Sunak who thus sees his initiative lasting. The British Prime Minister has made AI one of his hobby horses. At the event he also said that a previously announced investment of £100 million in two supercomputers was to be increased to 300 million.
While he welcomes the symbolic significance of the summit, Jean-Baptiste Bouzige, President and founder of Ekimetrics, a French AI specialist, thinks that the ball is above all in the United States’ court.
« With leading AI companies located across the Atlantic, Joe Biden has the greatest leverage. It is also rumored in the American tech community that responsibility for AI will become a key element for winning calls for tenders from next year: it will be impossible to work with Fortune 500 companies if we cannot prove that the data is processed properly. »
A vote at the UN to regulate autonomous weapons
Finally, the week was also marked by a vote in the United Nations General Assembly on Wednesday, in favor of a motion to control the use of autonomous weapons. This vote represents the first step in a process aimed at banning weapons without significant human control (e.g. a military drone that would act without human supervision), as well as proposing a framework for the development of autonomous weapons.
« It is time to adapt international law to technological developments which make it possible to develop weapons with an increasing degree of autonomy. Significant work has been carried out over the past ten years on this subject; it is now necessary to provide ourselves with a legal framework in the form of an international treaty signed by as many States as possible. This is essential to stem the risks of proliferation, it would send a strong signal to all stakeholders, particularly arms producers. », Notes Anne-Sophie Simpere, coordinator of the campaign against killer robots in France.
1699049736
#busy #week #events