A global push for AI ethics

The EU today published a set of ethical guidelines for “trustworthy AI” — a long wishlist of idealistic principles, many still technically out of reach, meant to keep unwanted harms from the powerful technology at bay.

Kaveh reports: It’s an early, earnest attempt to get countries to buy into general ethical principles. But without an enforcement tool, it is unlikely to result in safe AI.

The big picture: Up to now, AI development has largely been a free-for-all.

• The big playershave taken off on their own, ending up in familiar places: The U.S., favoring a hands-off approach, has left responsibility with Big Tech; Beijing imposes its own norms on Chinese companies; Europe has hewn a middle path.
• But there’s a growing realizationthat current unsupervised AI development is headed into dangerous territory. This has prodded experts to work toward shared norms for AI.

Why it matters: The driver’s seat of global AI policy is still empty, awaiting the country or organization that will set the rules for others to follow.

• Much rides on who takes control.A world with the U.K. at the helm would look very different from one with China chauffeuring.
• “Everyone wants to dictatewhat’s happening,” says Amy Webb, an NYU professor and founder of the Future Today Institute.

With today’s announcement, the EU has scored the first-mover advantage, says Chris Padilla, IBM’s VP for government and regulatory affairs, as it did with GDPR, the landmark privacy bill that has changed the way Big Tech does business in Europe.

• The new guidelinesare a far cry from GDPR’s strict rules.
• But that doesn’t meanthey’re useless, argues OpenAI Policy Director Jack Clark, who helped build an equivalent set of recommendations for the OECD. Governments will likely use these guides as templates for their own national policies, he tells Axios.
• Before regulations kick in,international standards and industry self-policing are especially important, says Charlotte Stanton, Silicon Valley director for the Carnegie Endowment for International Peace.

What’s next: The EU guidelines will soon have stiff competition.

• In the coming months,the OECD will come out with its own recommendations, with many similarities.
• S. and European officialsare also considering a proposal by Webb for an international body to oversee big tech companies, she tells Axios. A spokesperson for the White House Office of Science and Technology Policy would not confirm that it’s considering the proposal.
• AI pioneer Yoshua Bengiois pushing for binding ethical regulations as part of a new organization for governments, nonprofits, companies, and other experts.

In an interview with Nature, Bengio says, “Self-regulation is not going to work. Do you think that voluntary taxation works? It doesn’t.”

• Webb says much the same: “Developers are not incentivized at all to follow these guidelines,” she tells Axios, emphasizing that building in ethics would slow companies’ AI systems and perhaps make them less capable.
• What’s more, the growing buffet of ethics proposals may cause problems, Webb argues. A firm deciding between several international guidelines, its home country’s national policy, and recommendations from universities and nonprofits might end up doing nothing.

Go deeper:

• AI’s uneasy coming of age(Axios)
• Europe’s silver bullet in global AI battle: Ethics(Politico)
• China wants to shape the global future of artificial intelligence(Technology Review)

Πηγή: axios.com

 
• EU unveils ethics guidelines for artificial intelligence

The European Union presented ethics guidelines Monday as it seeks to promote its own artificial intelligence sector, which has fallen behind developments in China and the United States.

The European Commission, the bloc’s executive arm, unveiled a framework aimed at boosting trust in AI by ensuring, for example, data about EU citizens are not used to harm them.

“Ethical AI is a win-win proposition that can become a competitive advantage for Europe: being a leader of human-centric AI that people can trust,” Commission Vice President Andrus Ansip said.

The guidelines list seven key requirements for “trustworthy AI” established by independent experts consulted by the Commission.

Among them is one ensuring that data about citizens will not be used to harm them or to discriminate against them.

The measures also call for mechanisms to ensure accountability for AI systems and for AI algorithms to be secure and reliable enough to deal with errors or inconsistencies.

The Commission now aims to launch a pilot phase in which industry, research and public authorities test the list of key requirements.

It will also involve companies from other countries and international organisations.

The Commission aims to improve cooperation with “like-minded partners” such as Japan, Canada or Singapore and continue working with the G7 and G20 groups of leading economies.

The updated guidelines flow from the Commission’s AI strategy unveiled in April last year, which aimed to bring public and private investment in the sector to at least 20 billion euros annually over the next decade.

Europe is trying to catch up with both the US and China.

A study published last month showed that China is poised to overtake the United States in artificial intelligence with a surge in academic research on the key technology.

A burgeoning sector, AI is already used to recognise people in photos, filter unwanted content from online platforms and enable cars to drive themselves.

Πηγή: phys.org