Researchers' trust probably won't be regained without a more individual approach to this, though. We are a small and tightly knit community; there is not really a way to regain trust while ignoring damage to individuals in our community.
That Google can't see this is concerning. Also, losing researchers' trust has a contagious effect in the engineering recruiting pool by way of faculty opinions.
I'll work with individuals in the ITP community who happen to be at Google because they are my friends, and I won't tell students what to do because that's not my place and can hurt my friends. But not everyone agrees with that POV, social media exists, and so on.
I'll still strongly voice my opinions when relevant. And my opinion is that repairing damage to harmed individuals is necessary, otherwise Google has not regained my trust.
If you have an entire generation of faculty doing this, it is not going to be good for Google in the long run. Google should realize this and change their current approach. The current approach is not just cruel but also shortsighted.
I counter it and propose (by way of a public conversation with Timnit yesterday):
1. Google lets Jeff talk to the Ethical AI team directly and ask what they want; Jeff does that and acts on the advice.
2. Google lets Jeff talk to Timnit and Meg; Jeff does that, too.
1. Google lets Jeff talk to the Ethical AI team directly and ask what they want; Jeff does that and acts on the advice.
2. Google lets Jeff talk to Timnit and Meg; Jeff does that, too.
Much less shortsighted. Good for Timnit, Meg, Google, and Jeff, even if some initial pain precedes the benefits, like ripping off a bandaid.
The "Google lets" part is just based on educated guesses about how these things work in practice at companies. Most likely Jeff is acting within a system (say, according to advice from PR & legal), so nudging higher up for a change in the approach used seems ideal.
And I believe the reason this approach is not currently being taken is because of a misunderstanding higher up of how research communities work, and an underestimation of the damage of not taking such an approach.
Folks who make it high up in companies tend to have lawful D&D alignment, so it seems unlikely that asking for rulebreaking would succeed, and more likely that asking for a change in rules would succeed. There is a good case a change in rules is in Google's best interest.
So how could rules change?
I tried emailing @sundarpichai with a quite long and detailed and justified case on this, but I imagine he is busy and may not even read this email, let alone take it into account.
I tried emailing @sundarpichai with a quite long and detailed and justified case on this, but I imagine he is busy and may not even read this email, let alone take it into account.
What about the stakeholders? Likely they do not understand how research communities work, either. Is there a way to reach stakeholders to explain the harms of the current approach, and make a case for a different one?
What about other leaders within Google? Who understands the current organizational structure well; who plans these approaches? Is there anyone there open to learning more about how research communities work? I do not have answers to these questions, but I would like them.
Concrete harms of the current approach Google is taking extend well beyond Twitter. You'll find this discussion in every Slack channel: university Slack, global Slacks for faculty candidates. We discuss it one-on-one. Students ask us questions about it privately.
That this has not permeated stakeholder opinions enough to impact the stock price at all, or senior leadership enough to influence a change in strategy, shows a failure to understand the influence these conversations have and the dynamics of research communities.
How do we communicate this message effectively to people with influence over the strategy? This should be the focus, I think, of anyone who cares strongly about a just outcome for the impacted individuals, and I truly and strongly believe it is best for Google as well.
Feel free to circulate in order to gather ideas for this. Feel free to DM if uncomfortable strategizing publicly. Since I believe this is also in Google's best interest and in Jeff's best interest, and have already communicated this strategy to Sundar, I feel OK with public.
My understanding from public conversations with Timnit and Meg available on Twitter is also that the team has tried this sort of thing before and failed, so I think it will take more voices than just those of Ethical AI.
Faculty have so much power here, and can do this responsibly without any threats to Google just by communicating how our community works to people with a say in strategy. But Googlers and ex-Googlers would probably need to help us find those people.
I guess, since I want to operate with complete transparency, but I continue to feel strongly invested in this because I care deeply about the community, I hope you all see this: @timnitGebru @mmitchell_ai @JeffDean @sundarpichai @GoogleAI
Have a good afternoon.
Have a good afternoon.
Ah, and faculty with aforementioned power, I hope you see this too. But I think that is like, hundreds of my followers, at the very least, too much to tag for now.
Of course long-term there should be higher-level systemic changes at the level of regulation. This is not mutually exclusive with making the case for individual justice, though.
@threadreaderapp unroll
• • •
Missing some Tweet in this thread? You can try to
force a refresh
