Hacker News Reader

Comment by mldrama2020 | Hacker News Reader

mldrama2020

She is a loose cannon and a big bully. I am surprised Google waited this long to let her go. In Yann vs Timnit debate, when Yann asked apology for the escalating the situation, she threw a document and asked him to apologize and take responsibilities like that in the document.

View on HN

I've seen this exact text in at least five responses to this and similar articles, all without any substantive information. Sad and embarrassing, HN.

https://syncedreview.com/2020/06/30/yann-lecun-quits-twitter...

The "debate" was a Twitter spat, dragging AI&Ethics research down to the level of, well, Twitter. This is especially sad because everyone invovedt has proven they can do better by writing legit research papers.

Indeed, these arguments on Twitter aren't a lot of fun to read and seldom go anywhere useful. That said, this certainly strikes me as one of the more restrained such exchanges. In my opinion, Yann LeCun wasn't being bullied here.

> I've seen this exact text in at least five responses to this and similar articles, all without any substantive information. Sad and embarrassing, HN.

She is giving bits and pieces as information. With already available information, she asked Google to reveal the name of internal reviewers. If not she threatened to quit, Google said bye to her.

If someone was preventing my paper from being published, pretty much at the deadline, I would also like to know who they were and address their concerns directly. If everyone at Google is a grown adult, I think they should be willing to defend their last-minute issues with this person's paper directly.

Why should they need to shroud themselves in secrecy?

> [...] defend their last-minute issues [...]

Jeff Dean's response claims that the paper was "only shared with a day’s notice before its deadline", when Google "require two weeks" for their internal review process. So the issues were last-minute, but only because the paper was.

> Why should they need to shroud themselves in secrecy?

Anonymity is common in paper reviews, to prevent retribution and help keep it objective.

It is no secret. The paper attacks BERT and other LMs, which in turn affects Google's bottom line. If you want to write such a paper, go and do it in Academia not in Google's payroll.

You do not have freedom to publish anything you want in industry. That's just common sense, mate.

I have to disagree, isn't this exactly the sort of thing Google hired them for? Wikipedia lists their job title as "technical co-lead of the Ethical Artificial Intelligence Team at Google."

"She is a loose cannon and a big bully."

I do not know her personally. The overwhelming support for her in the ML community appears to show the opposite. She appears to be highly regarded and well-respected, personally and professionally.

From reading Jeff Dean phrase her request, this was the sense of what he and the company want to illustrate. But reading his assistant’s termination email and Gebru’s emails to management and subsequently to allies (for which she was quoted as not being worthy of a manager), the real issue is the company feels she’s too independent and threatened by her lack of submissive attitude. This has spoken more about the company’s attitude toward culture and perception, elucidating the very points in Gebru’s email to Brain allies that the company doesn’t care about opinions or humanity. Ironically, the very reason she is in her role to help with as an ethicist - which also puts into question how they see this role not as innovative but as a shill. Gebru’s mistake is she worked for the wrong employer.

Reply on news.ycombinator.com