What is 'responsible' in Responsible AI?

By Os Keyes

Google recently announced a new advisory board on responsible AI, and it’s a nice demonstration of precisely how fucked up things are in AI ethics right now.

The biggest and most obvious WTF is the presence of the President of the Heritage Foundation, an American far-right thinktank whose politics are eye-wateringly terrifying. From xenophobic “build the wall” rhetoric to climate change denial, they have it all, including anti-abortion stances and positions in favour of discriminating against gay people.

Their primary horror from my perspective is their virulent transphobia. Seeking to treat trans rights as a wedge issue to split LGBT rights overall, they’ve funded the British TERF movement and actively campaigned against trans-inclusive legislation - campaigns that their President, in particular, publicly acknowledges and supports.

Quite why Google thinks they add to a conversation about responsibility is unclear, although Joanna Bryson, a professor also appointed to the panel who thinks that slavery is a good rights model for robot ethics, accidentally made the subtext text when she noted that they wanted to be able to reach the American far-right. Bryson happens to endorse this; indeed, she’d work with war criminals if it got something done.

But whatever the reason, the Heritage Foundation’s presence - and the attitudes expressed by the self-proclaimed “ethicists” appointed from out of academia - are damning to both Google and the academics appointed. If you are on this panel, you are communicating that you consider transphobia, homophobia and racism acceptable attitudes in a colleague if the status bump is good enough. There is nothing ethical or responsible in that: instead, it marks the absolute absence of responsibility: it marks you saying that you’d trade your queer or immigrant students’ comfort, safety and certainty for an unpaid google role. Every day the Heritage Foundation’s president remains on the panel, and people like Bryson, Luciano Floridi and other academics do not quit people are watching and taking note. And in the long term, this is worse for Google’s ethical reputation, and the reputations of appointees, than no advisory panel at all.