White House leads think-tank on AI

The US government has joined forces with a group of academics, industry experts and law experts to look at how artifical intelligence interacts with law and policy. In the first of four gatherings around the US, White House Edward Felten, White House Deputy US Chief Technology Officer, said that there were a range of policy challenges which needed to be considered around AI, its risks and challenges and how it would impact on the future. Some of these challenges would be more challenging over time, he said.  From setting up a regulatory regime to applying the power of AI to help the public, ensuring that technology remains safe and impacting the economy, a collaborative approach was needed. The problems were too big to be addresed by technologists and lawyers alone or any one community.  Mr Felden said that the government had put together a group to look at all these problems in keeping with President Obama's focus on science and technology which reflected the diversity of the US.

Challenges

Big data expert Kate Crawford, Principal Researcher at Microsoft Research New York City and a Visiting Professor at MIT's Center for Civic Media, said privacy was one such issue - and that she worried that a limit had been reached over privacy. With the mass collection of behavoural data, issues were inevitable over ownership and privacy. However, she believed this should be thought about structurally and how it impacted entire groups, not just individuals. Another issue which was on the cards was transparency around algorithms. A recent study on risk assessments in criminal sentencing revealed that there was disparity between African Americans compared to white Americans with the former labelled as higher risk when algorithms were used.  This impacted on sentencing. Ms Crawford pointed out that this highlighted issues over transparency and algorithms which were particularly serious when built into criminal justicea and education systems. Jack Balkin, Knight Professor of Constitutional Law and the First Amendment at Yale Law School, said that whilst professional duties may be de-professionalised through AI, a new set of norms could arise which would see re-professionalisation. 

Other issues

Other issues which were discussed included the growth of surveillance capital. Deirdre Mulligan, Associate Professor in the School of Information at UC Berkeley, said it was a real opportunity to get to grips with the issues - as it was early days. Edward Felten said it was important to make sure regulation could adapt - and that this was the preferred way rather than starting from scratch. 'We are nowhere near talking about broad regulation of AI,'he said adding that there was robust legislation in place across a range of areas.

Defence 

One observation was that the Department of Defence had put AI at the centre of its defence policy which led to the question - could there be AI for good and AI for bad. However, Mr Felton said that whilst the difficulty of autonomous weapons was a real one, most defence revolved around logistics and intelligence. He pointed out that all treaties apply to military and that these called for appropriate levels of human control.

Roadshow

The event, at the University of Washington School of Law, was co-hosted by the White House and UW’s Tech Policy Lab and is the first of four organised to look at different areas in which AI will make an impact.