Clumsy AI rhetoric is bad for everyone

Be European The Alliance Committee issued its Regulatory recommendations In the field of artificial intelligence last month, many U.S. policy circles cheered. Their praise is based at least in part on the fact that the most powerful democracies in the world have not adequately regulated AI and other emerging technologies, and the document marks a step forward.However, in most cases proposal The response to this highlights the confusing rhetoric about AI in democracies.

Over the past decade, high-level goals for regulating AI have often conflicted with the details of the specification proposal, and in both cases, what the final state should be has not been well articulated. Although there may be differences between countries, consistent and meaningful progress in the development of democratic AI regulations with international appeal begins with the resolution of many contradictions and subtle features in the discourse.

The European Commission has touted its proposal as a milestone in AI regulations. Executive Vice President Margrethe Vestager (Margrethe Vestager) Say After the release, “We think it is urgent. We are the first person on the planet to propose this legal framework.” Another commissioner Thierry Breton (Thierry Breton) Say These proposals “are aimed at strengthening Europe’s position as a hub of global AI excellence from the laboratory to the market, ensuring that European AI respects our values ​​and rules, and realizing the potential of AI in the industrial field.”

This is certainly better than the stagnation of road rules for companies, government agencies and other agencies in many countries, especially the United States. Despite minimal supervision and accountability, AI has been widely used in the European Union. Surveillance in Athens or Operating buses In Malaga, Spain.

But the EU regulations are considered “leading” only because First It only covers up many problems of the proposal. This leap in rhetoric is one of the first challenges facing the democratic AI strategy.

among all the In the 108-page proposal, “norm” is particularly important.It reads: “For law enforcement purposes, the use of artificial intelligence systems to perform’real-time’ remote biometric identification of natural persons in public places is considered to particularly violate the rights and freedoms of the persons concerned,” it will affect private lives and “arouse people’s constant The feeling of being under surveillance” and “indirectly discourages the exercise of freedom of assembly and other basic rights.” At first glance, these words may mean concern of a lot of Activist with technology Ethicist The hazards of face recognition may bring serious large-scale surveillance risks to marginalized communities.

The committee then stated: “Therefore, the use of these systems for law enforcement purposes should be prohibited.” However, this would allow for exceptions “in three well-listed and narrowly defined situations”. This is where the loopholes come into play.

Exceptions include “circumstances involving the search for potential criminal victims, including missing children; certain threats to the life or personal safety of natural persons or terrorist attacks; and the detection, location, identification or prosecution of offenders or criminal suspects.” For all Described as a “narrowly defined” scenario, this language provides countless reasons for law enforcement officers to deploy facial recognition as needed. For example, allowing it to be used in the “perpetrator or suspect” of a criminal offence can accurately allow activists to be wary of racial and sexist facial recognition algorithms for a long time.

EU’s privacy regulator, European data protection director, Pounce quickly here. The EDP​​S statement said: “In view of the fact that remote biometrics may lead to unprecedented development, and artificial intelligence may bring tremendous depth to unprecedented development and the risk of non-democratic intrusion into the private life of individuals, a more rigorous approach must be adopted. ,” the EDPS statement said. Sarah Chander of the non-profit organization “European Digital Rights” description The proposal to the edge is “the single board for protecting basic rights.” Others have noticed how these exceptions reflect U.S. legislation, which on the surface seems to restrict the use of facial recognition, but in fact there are many broad breakthroughs.