We read the paper that forced Timnit Gebru out of Google • Hacker News Board

Re: We read the paper that forced Timnit Gebru out of Google

#201

Post by choeger » Sat, Dec 05, 2020, 10:47 AM UTC

While I will not dive into the extremes here, this is yet another battle in the fight over language, right? The most important point of criticism seems to be that the language models are "unethical", i.e., not subject to control by a particular political standard.

Am I the only one that finds this line of argumentation highly troubling? The idea that once you do something with language there should be someone proactively controlling you? Should it not always be the output that will be judged by the public?

Re: We read the paper that forced Timnit Gebru out of Google

#202

Post by the-dude » Sat, Dec 05, 2020, 10:47 AM UTC

Jabbles wrote: post #187

bartread wrote: post #158

Good grief - this terrible clickbaity headline writing: "We read the paper that forced Timnit Gebru out of Google. Here’s what it says" The paper DID NOT force her out of Google. Her subsequent behaviour - submitting without approval, rant, ultimatum, and resignation - did. And she wasn't "forced out": she resigned of her own volition. She could have chosen to make improvements to the paper based on the feedback she…

I don't see how you can possibly be that confident about the truth of what happened. Unless you have non-public information, then there is still considerable uncertainty. Timnit says she was fired: https://twitter.com/timnitGebru/status/1334352694664957952 Jeff Dean's email says the article was "approved", though clearly not approved enough: https://www.platformer.news/p/the-withering-email-that-got-a... Other resear…

And further down that thread:

> I need to be very careful what I say so let me be clear. They can come after me. No one told me that I was fired.

Re: We read the paper that forced Timnit Gebru out of Google

#203

Post by threeseed » Sat, Dec 05, 2020, 10:48 AM UTC

StavrosK wrote: post #174

foota wrote: post #139

Earlier quoted context omitted.

Her previous papers look pretty serious to me, along with her resume. I see nothing to suggest she's not qualified. https://ai.stanford.edu/~tgebru/

What do you mean by pretty serious? I looked at three, two are about better documenting ML models/data, and one is about inferring the political leanings of an area by what type of cars are parked in it. None seem particularly groundbreaking to me...

Her work on identifying bias in facial recognition has won quite a number of awards.

Ground-breaking might be a strong word but it does move the industry forward.

Re: We read the paper that forced Timnit Gebru out of Google

#204

Post by saagarjha » Sat, Dec 05, 2020, 10:48 AM UTC

ardit33 wrote: post #125

Ok, so basically she is a bullshit social engineer, masquerading as an 'AI ethics researcher' "Gebru’s draft paper points out that the sheer resources required to build and sustain such large AI models means they tend to benefit wealthy organizations, while climate change hits marginalized communities hardest. “It is past time for researchers to prioritize energy efficiency and cost to reduce negative environmental i…

I am honestly disappointed to see this kind of response at the top of comments here. Every time we have some sort of social issue come up people here come in to pick apart the person involved, and I think it’s one of the bigger reasons why some maintain a generally negative view of Hacker News. Can you imagine being Timnit and stopping by this thread, where the top comment, in colorful language, reduces you to a social activist fraud? A comment coming from someone who hasn’t even read the full paper? One that ignores the job you were hiring to do (which seems precisely to draw out the social issues of AI) and lambasts it?

At you the commenter yourself: you seem to be exceptionally harsh on her based on the most minimal information that you’ve purposes to fit your claim of this being “useless things that we shouldn’t care about”. Do you know how much energy goes into training a model? I certain didn’t. Do you think that showing that AI is picking up discriminatory language is something worth looking into? I certainly think having someone look into it would be useful, yes. And Google apparently thought so too. Claiming that these things are idiotic and attacking the author of this paper for bad faith and censure is…very extreme.

Re: We read the paper that forced Timnit Gebru out of Google

#205

Post by jawarner » Sat, Dec 05, 2020, 10:50 AM UTC

ardit33 wrote: post #125

Ok, so basically she is a bullshit social engineer, masquerading as an 'AI ethics researcher' "Gebru’s draft paper points out that the sheer resources required to build and sustain such large AI models means they tend to benefit wealthy organizations, while climate change hits marginalized communities hardest. “It is past time for researchers to prioritize energy efficiency and cost to reduce negative environmental i…

Google’s algorithms will social engineer whether Google likes it or not. They might as well be intentional about it.

Re: We read the paper that forced Timnit Gebru out of Google

#206

Post by threeseed » Sat, Dec 05, 2020, 10:53 AM UTC

rualca wrote: post #199

adar wrote: post #190

Earlier quoted context omitted.

> I didn't noticed the OP making any remark on the author's qualifications. Did you not read the first sentence of OP's post? "Ok, so basically she is a bullshit social engineer, masquerading as an 'AI ethics researcher'"

If you take a step back and read the snippet you cherry-picked, I believe you'll be able to understand that it referred to how the author is trying to pass opinions on social engineering under the guise of research on AI ethics. That does not mean at all that the author lacks qualifications.

When you accuse someone of pretending to be an "AI ethics researcher" you are stating that they lack the qualifications to hold that title.

Pretty obtuse to suggest otherwise.

Re: We read the paper that forced Timnit Gebru out of Google

#207

Post by visarga » Sat, Dec 05, 2020, 10:53 AM UTC

MperorM wrote: post #175

_cs2017_ wrote: post #116

Despite what both sides claim, IMHO papers like this are primarily PR, ideology, and politics, rather than science and technology. This is akin to a speech writer working for a politician, writing a piece that disagrees with the party platform, and refusing to fix it when asked.

To her defense, she was working on ethics in AI. What ethics to adopt inherently comes down to ideology. While I don't really understand the emissions argument, the rest strikes me as very defensible. If the best language models need giant datasets to excel, it is very difficult to ensure your AI is trained on reasonable data. I wouldn't want an AI therapist to be trained on 4chan. Now obviously nobody would be that…

You train your LM on web crawl data, but also train a 4chan classifier, then you condition your LM not to generate in 4chan style. GPT-3 got a similar chaperone classifier for offensive speech. It's like knowing swear words but choosing not to use them. You could also condition a general LM to bias and debias its outputs as you like.

Re: We read the paper that forced Timnit Gebru out of Google

#208

Post by bartread » Sat, Dec 05, 2020, 10:54 AM UTC

Jabbles wrote: post #187

bartread wrote: post #158

Good grief - this terrible clickbaity headline writing: "We read the paper that forced Timnit Gebru out of Google. Here’s what it says" The paper DID NOT force her out of Google. Her subsequent behaviour - submitting without approval, rant, ultimatum, and resignation - did. And she wasn't "forced out": she resigned of her own volition. She could have chosen to make improvements to the paper based on the feedback she…

I don't see how you can possibly be that confident about the truth of what happened. Unless you have non-public information, then there is still considerable uncertainty. Timnit says she was fired: https://twitter.com/timnitGebru/status/1334352694664957952 Jeff Dean's email says the article was "approved", though clearly not approved enough: https://www.platformer.news/p/the-withering-email-that-got-a... Other resear…

Timnit also shared a number of tweets, which you can easily access via the one you referenced, where she quotes from the email that Google sent her in response to her ultimatum.

They accepted her resignation and brought her finish date forward. They cited her message to the brain group as reason for doing so (without the resignation I suspect she may have faced some disciplinary action, though whether it would have gone as far as firing I don't know).

They clearly weren't happy with the content but, beyond that, if somebody is pissed off enough to write that the kind of message Timnit did then, by making them work their notice period, you only invite them to cause more trouble whilst they're still part of the organisation. You therefore bring forward their leaving date and make their resignation effective immediately.

When you do this it's about mitigating risk to the organisation. Commonly I've seen it done with salespeople in certain sectors, where when they resign they are escorted from the premises and access is revoked as part of minimising the risks that they'll take clients with them to their next role (particularly if they'll be working for a competitor). Still, any situation in which continuing to have an employee around represents a significant risk to the organisation is one in which you might ask them to leave immediately.

Google already had a mess to try to clean up with the brain group as a result of Timnit's message to that group. They probably didn't want any new messes to deal with, so they asked her to leave immediately to mitigate that risk.

Btw, I'm not advocating for Google here: I'm just looking at this in terms of, "What would I as a manager do in a circumstance where an employee has set out conditions of an ultimatum for their continued employment that I am unable or unwilling to meet?"

Re: We read the paper that forced Timnit Gebru out of Google

#209

Post by StavrosK » Sat, Dec 05, 2020, 10:56 AM UTC

threeseed wrote: post #203

StavrosK wrote: post #174

Earlier quoted context omitted.

What do you mean by pretty serious? I looked at three, two are about better documenting ML models/data, and one is about inferring the political leanings of an area by what type of cars are parked in it. None seem particularly groundbreaking to me...

Her work on identifying bias in facial recognition has won quite a number of awards. Ground-breaking might be a strong word but it does move the industry forward.

Do you have a link to that paper? I'd like to read it, though I wasn't thrilled by the way she handled (or didn't) Yann LeCun.

Re: We read the paper that forced Timnit Gebru out of Google

#210

Post by rualca » Sat, Dec 05, 2020, 10:56 AM UTC

threeseed wrote: post #206

rualca wrote: post #199

Earlier quoted context omitted.

If you take a step back and read the snippet you cherry-picked, I believe you'll be able to understand that it referred to how the author is trying to pass opinions on social engineering under the guise of research on AI ethics. That does not mean at all that the author lacks qualifications.

When you accuse someone of pretending to be an "AI ethics researcher" you are stating that they lack the qualifications to hold that title. Pretty obtuse to suggest otherwise.

> When you accuse someone of pretending to be an "AI ethics researcher" (...)

You would do well to work on your literacy and reading comprehension skills.

To start off, be aware that "a bullshit social engineer masquerading as a AI research" does not, by any means, mean the author is "pretending to be an AI researcher".