AI Safety Summit: AI needs rules, but who will make them?

»I completely agree that AGI risks are a distraction and I take away the concern about capture [der Politik] by companies seriously. But I’m also starting to worry that simply trying to focus on these risks disproportionately favors corporate interests at the expense of people,” says Margaret Mitchell, senior ethics researcher at AI company Hugging Face. (The company was present at the Bletchley Park summit, but Mitchell himself attended a forum in the US hosted by Senator Chuck Schumer of New York State).

»People should be the focus, not technology«Melanie Mitchell, Ethikforscherin

“When regulating AI, the focus should be on people, not technology,” says Mitchell. “And that means we should focus less on what this technology could do and how we categorize it – and more on how we can protect people.” Mitchell’s reluctance to take the risk-based approach stems in part from that numerous companies at the UK summit and similar events in the first week of November were willing to join this approach. “That immediately set off alarm bells for me,” says the ethics researcher, adding that she took a similar stance at Schumer’s forum in New York.

A rights-based approach to AI

Mitchell advocates for a rights-based approach to regulating AI rather than a risk-based one. This also applies to Chinasa T. Okolo, a fellow at the Brookings Institution who attended the event in Great Britain. “The main discussions at the summit revolve around the risks that ‘frontier models’ pose to society,” she says, “but they ignore the damage that AI does to data labelers [beim Sichten und Labeln der Trainingsdaten] – those workers who are fundamental to AI development.«

When focusing on human rights, the discussion takes place in an area that politicians and regulators are more familiar with than with technology. Mitchell believes this would help lawmakers confidently craft laws to protect as many people to whom artificial intelligence poses a potential threat. It could also represent a compromise for the tech companies so anxious to protect their established positions – and their billions of dollars in investments. “By focusing policy on rights and goals, you can combine top-down regulation, which governments are best at, with bottom-up regulation, which developers are best at,” concludes Mitchell.

The post first appeared on

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top