Google is investigating the actions of another top AI ethicist
1–10 of 125 posts
Re: Google is investigating the actions of another top AI ethicist
#2> In this instance, yesterday our systems detected that an account had exfiltrated thousands of files and shared them with multiple external accounts.
I mean, come on. There’s no way someone in a leadership position would legitimately think that’s okay. I suppose to some degree we should reserve judgment until hearing her side of the story - maybe this was just unambiguous evidence destined for her lawyer and the EEOC - but it’s hard to avoid some pretty negative conclusions about the prevailing culture in this group.
Re: Google is investigating the actions of another top AI ethicist
#3I was disappointed to find out that AI ethicists at Google aren't LessWrong style AI ethicists. I've just read one Google AI ethics paper in it's entirety but it seemed pretty non-insightful and not actionable.
That said, if Google is "investigating" her because she was trying to find evidence of discrimination or bad treatment of Dr. Gebru, that seems borderline criminal behavior. (Assuming her attempts at trying to find evidence did not involve exfiltrating secret information).
Re: Google is investigating the actions of another top AI ethicist
#4Another perspective on this drama to me is what the official expectation of a team labelled "Ethical AI" is/was. I doubt the only focus was to prevent discrimination by sex or race,
For example, there's also prevention of mass manipulation (AI learns human psychology, $BAD_ACTOR uses that to manipulates masses) and AI based profile generation of individuals with the goal of creating leverage for blackmail/extortion/whatever. Or, how would one form a controlling body of unethical usage in AI in such areas. There are quite a few aspects never mentioned in these discussions, and a search via google scholar turns up lots of the sex/gender/race papers, but none of the others (at least on the first few pages)-
Re: Google is investigating the actions of another top AI ethicist
#5Do Google corporate email accounts allow IMAP sync, like gmail accounts? I mean fully-replicated BODY.RFC822 sync, not just headers.
I have to assume not, otherwise everybody using this feature would trigger the same exfiltration-lockout.
Re: Google is investigating the actions of another top AI ethicist
#6Another perspective on this drama to me is what the official expectation of a team labelled "Ethical AI" is/was. I doubt the only focus was to prevent discrimination by sex or race, For example, there's also prevention of mass manipulation (AI learns human psychology, $BAD_ACTOR uses that to manipulates masses) and AI based profile generation of individuals with the goal of creating leverage for blackmail/extortion/w…
Politics is a means to an end. Investigating power abuse through one's own technology would be self-harm from the POV of a mega cooperation.
Re: Google is investigating the actions of another top AI ethicist
#7Do Google corporate email accounts allow IMAP sync, like gmail accounts? I mean fully-replicated BODY.RFC822 sync, not just headers. I have to assume not, otherwise everybody using this feature would trigger the same exfiltration-lockout.
Or instead of answering my question people could just downvote me. I guess.
Legitimately curious here. "Download all my mail from the mail server" is something I do multiple times every day. I guess my workflow would be verboten in corporate environments? Is it really such an odd behavior that they have an automated check for this, and it isn't throwing false positives every hour? Or is this one of those things they tell you at the "onboarding" like "hey, you might download all your personal gmail via IMAP regularly, but don't do that with your corporate email account or you'll get a phone call from security".
Maybe I'm just out of touch with how corporate email works. It's been a decade since I was an employee of a publicly traded company.
Re: Google is investigating the actions of another top AI ethicist
#8“In this instance, yesterday our systems detected that an account had exfiltrated thousands of files and shared them with multiple external accounts.”
If true, that is damning, and would demonstrate once again that being an “ethics researcher” does not mean that you are are any more ethical than the average person. It just means you are more interested in the subject.
As a side note, I wish this field was more interested in meta-ethics than it is in forcing machines to abide by the personal ethics of the humans involved.
Re: Google is investigating the actions of another top AI ethicist
#9I was disappointed to find out that AI ethicists at Google aren't LessWrong style AI ethicists. I've just read one Google AI ethics paper in it's entirety but it seemed pretty non-insightful and not actionable. That said, if Google is "investigating" her because she was trying to find evidence of discrimination or bad treatment of Dr. Gebru, that seems borderline criminal behavior. (Assuming her attempts at trying to…
I think sending it to external accounts probably triggered this, there is no way this would not have raised a red flag in any company. I mean, Alphabet is the same company where a person took self driving documents with him to start a new company.
Hopefully, if it was just normal conversations and after identities had been redacted, their access is restored. Still, analysing this information on their work machine itself and then sending out a summary seems like a better course of action than this.
Re: Google is investigating the actions of another top AI ethicist
#10Do Google corporate email accounts allow IMAP sync, like gmail accounts? I mean fully-replicated BODY.RFC822 sync, not just headers. I have to assume not, otherwise everybody using this feature would trigger the same exfiltration-lockout.
Or instead of answering my question people could just downvote me. I guess. Legitimately curious here. "Download all my mail from the mail server" is something I do multiple times every day. I guess my workflow would be verboten in corporate environments? Is it really such an odd behavior that they have an automated check for this, and it isn't throwing false positives every hour? Or is this one of those things they…
In general: if it’s on a corporate server, it’s not your email, it’s corporate IP.