Terah Lyons’s job is making sure that AI behaves itself.
Copyright by www.wired.co.uk
In October 2017 Lyons, a former White House advisor on technology and machine intelligence, started her tenure as an executive director at Partnership on AI.
Launched the previous year by a squad of tech titans – Amazon, Facebook, Google, DeepMind, Microsoft, IBM – by then the organisation had already grown to include about 50 members, including non-profits, charities, and universities. (As of September 2019, the Partnership had grown to encompass 89 stakeholders from 13 countries.)
The Partnership’s main aim was to work out a rulebook for the conscientious use of artificial intelligence, and Lyons had taken the helm exactly at the time when a vast chunk of the public had become convinced that thrashing out some rules to rein in Big Tech was more urgent than ever.
“The partnership was born into a world in which AI was at the height of the public narrative. And the scrutiny of the technology industry that comes alongside that is a necessary part of a public movement to understand how technology is shaping our lives,” Lyons says. “It was really productive to have a forcing function for that discussion.”
Since Lyons stepped in, the organisation has produced several studies, reports, and recommendations on hot-button topics such as algorithmic decisions in the criminal justice system – about which the Partnership proposed a set of standards that, while unenforceable, might help companies developing predictive policing tools ask themselves the right questions. Other thematic projects are in the works.
And Lyons thinks there is some intrinsic value in bringing together entities from disparate corners of the globe to make AI safe, fair, and conducive to the social good. The model hearkens back to what has historically occurred in other critical technological fields. “This happened, for instance, with nuclear energy – with coalitions of competing organisations banding together,” Lyons says.
That can help build bridges across geopolitical divides – as testified by the presence of several Chinese companies among the partners. “But what’s really different about the Partnership on AI is that we really are a multi stakeholder entity,” Lyons explains. “It’s not just about the tech behemoths working hand-in-hand, but also about the voices of advocates and institutions that are pushing back to protect the public interest.”