Last week, Google announced that it had formed an AI Ethics committee to assist with the direction of its intelligent technology. This week came an update that few expected – the panel has already been dissolved.
Google has claimed that the committee “can't function” as it wanted. Although, many believe that the reason for the sudden collapse of the panel is due to the negative reaction from Google staff and the public to one of its key members.
Google has claimed that it will continue to research ethics in AI, though through alternative means.
What Has the AI Ethics Committee Been Disbanded?
Shortly after Google's announcement of its creation, the Advanced Technology External Advisory Council (ATEAC) came in for criticism. The backlash came from all fronts – the public and staff at Google – for some of the representatives that the company had chosen.
The focus of much of this disagreement was on Kay Coles James, the president of the Heritage Foundation – a conservative think tank. Questions were raised about her suitability as a representative on the panel, with a petition started by Google employees that she be removed from the panel due to her “anti-trans, anti-LGBTQ and anti-immigrant” views. The petition was signed by more than 2,500 staff, before being opened up to the public.
An example of the views held by Kay Coles James that upset many at Google and the public
Shortly after this, another member of the panel, Alessandro Acquisti stepped down from council, stating on Twitter that “While I'm devoted to research grappling with key ethical issues and fairness, rights and inclusion in AI, I don't believe this is the right forum for me to engage in this important work'.
Another member, Joanna Bryson, Associate Professor in the Department of Computer Science at the University of Bath, was questioned about her involvement with the group based on Kay Coles James' link on Twitter, and replied “Believe it or not, I know worse about one of the other people“. Bryson stated that she had questioned Google on its right-leaning associations in the past, and said that the company had replied that “it need[ed] diversity in order to be convincing to society broadly, e.g. the GOP“.
In a statement given to Vox, a Google spokesperson confirmed that ATEAC had been dissolved, stating “ATEAC can't function as we wanted”. However, it did not pinpoint the exact reason why, nor make any references to Kay Cole James or the Google staff petition.
What Were the Goals of the Ethics Committee?
As we reported last week, ATEAC has been tasked with shaping Google's AI policies and resolving complex ethical decision, with facial recognition and machine learning among the topics to be discussed.
Their work was to complement Google's “AI Principles”, a set of rules that the company established last year as it forged ahead in its AI research. Some have stated that these were a direct result of the company's decision to not renew its contract with the Pentagon to supply AI drones. Its prior involvement with the military caused ire among Google staff, who felt that it went against the company values.
The seven principles include creating AI that benefits society, avoiding creating unfair bias, being accountable, and being built and tested for safety.
Has Google Now Abandoned AI Ethics?
In its short statement on the disbandment of ATEAC, Google stated that it was “going back to the drawing board”, and that it would continue to highlight the issues raised by AI. But, Google has said it would “find different ways of getting outside opinions on these topics”.
What form the next iteration of Google's AI ethics plans takes remains to be seen. But, its statement suggests that it won't be another panel. The problem that the company faces is that if it truly is looking for a diverse group of policy makers from both the right and the left, it will always struggle to have them see eye to eye. While Google may be able to include both sides in its debates, it can't appease them both, too.
Skeptics had claimed that the ATEAC was limited in its ambitions, and that meeting four times a year wouldn't allow for much progress on such a cutting edge issue that will only continue to become more prevalent. Whatever its next move, Google is sure to tread carefully.
