It is surreal to watch the talk for our paper by @mcmillan_majora and @emilymbender at @FAccTConference. Never imagined what transpired after we decided to collaborate on this paper. Besides @mmitchell_ai, @blahtino @vinodkpg & @benhutchinson would also have been co-authors. 1/5
This paper already seems to be making an impact, so its extra disheartening that those who spent so much time contributing to it, and without whom it wouldn't have been what it is, aren't allowed to receive credit for it. Every single author was integral to the paper. 2/5
I just can't believe that this was the result of simply writing this paper. Honestly, just sitting here at a conference watching a talk, and waiting for Q&A and reflecting a tiny bit, I can't believe this was the result months after I first sent Emily this message: 3/5
"Hi Emily, I'm wondering if you've written something regarding ethical considerations of large language models or something you could recommend from others?...and if you haven't written something yet I would by customer #1 of anything you write on this end" 4/5
& her answer: "I can kind of see a paper taking shape here, maybe not saying that large language models are bad, but rather using large language models as a case study for ethical pitfalls and what can be done better.
Would you be interested in co-authoring such a thing?" 5/5

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Timnit Gebru

Timnit Gebru Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @timnitGebru

11 Mar
"Here, we detail the intensity and ferocity of this abuse, and demonstrate how it is designed not only to vilify a journalism icon, but to discredit journalism itself, and shatter public trust in facts...."
"These attacks have also created an enabling environment for Ressa’s persecution and prosecution in the Philippines. Now, her life is at risk and she faces the prospect of decades in jail, proving that there is nothing virtual about online violence."
"Despite fledgling attempts to address hate speech within the social media ecosystem, the platforms are the vectors which facilitate these attacks, creating an enabling environment for the State-led legal harassment of Ressa."
Read 4 tweets
11 Mar
@schrep @_KarenHao I would not respond like this to Karen's extensive, well thought out piece. Those of use at the forefront of this field, are not "less motivated" after reading the piece.

The issue is that we have fought hard to convince people in power, such as yourself who...
@schrep @_KarenHao have not faced the very real consequences of misinformation and all other issues that have been fueling all sorts of issues. Most recently, crimes against humanity in Ethiopia. @L_badikho has even been studying this in the Moroccan setting, ON HIS OWN TIME not because its fun...
@schrep @_KarenHao @L_badikho but because his life & the lives of many others are in danger & the shallow understanding of FB executives & lack of will to put required resources, makes it necessary for him to do this. Its impossible to convince those in power of the gravity because they're not the ones dying.
Read 6 tweets
11 Mar
cnn.com/2021/03/11/tec…. In depth article by @rachelmetz

"She responded to Bender that she was trying to get Google to consider the ethical implications of large language models."

Well. Thats how THAT went.

I didn't "exit" "leave" or "depart" tho in so much as I was fired.
"With very few laws regulating AI in the United States, companies and academic institutions often make their own rules about what is and isn't okay when developing increasingly powerful software. "
In a statement, Google said it has hundreds of people working on responsible AI, and has produced more than 200 publications related to building responsible AI in the past year

Thats their MO any time they fire someone, they like to make them look like they are JUST ONE of many
Read 17 tweets
9 Mar
The damage Jeff Dean has done with his gaslighting email & insinuating we were trying to dox anonymous reviewers in some sort of review process because we're not researchers who can take criticism: 1st Hacker News comment about this NBC article: 1\4
nbcnews.com/tech/tech-news…
"Important to remember where these allegations originate from. Timnit Gebru was fired from Google after threatening to quit unless Google reveled the identities of the reviewers of her paper that was not received well while accusing them of sex/race based discrimination."
2/4
news.ycombinator.com/item?id=263973…
"You can find some of the experiences of her from previous colleagues. You'll find the toxicity that she used in her time there and how she is picking the race/sex card on every situation. This is likely a continuation of that." 3/4
Read 6 tweets
9 Mar
"But after months went by with no meaningful update from the EEOC, I decided to pursue a lawsuit, even though plenty of attorneys advised me to leave it alone. “Google will eat you alive,” one said. "
"During the precious few hours of the day when I’m not working or caring for my children, I’m mired in the details of my legal battle. And, unlike Google, I don’t have a full staff working on my case. My entire life has been opened up for analysis and scrutiny:"
"Google has subpoenaed employment records from my subsequent jobs and, because I’m asking for emotional damages, they also get access to some of the most intimate details of my life, like my medical records. "
Read 5 tweets
26 Feb
One of the things Jeff Dean keeps on doing is thrashing this paper: arxiv.org/abs/1906.02243 by @strubell Ananya Ganesh & Andrew McCallum. Andrew MacCallum even has a Google affiliation from what I understand, at least he did when I was there. 1\4
This is one of the first papers where researchers estimate energy consumption in THEIR OWN research and development process of models THEY develop. They write down what process they used and how they made these estimates. Why someone wants to harp on this work is beyond me. 2\4
On the other hand, Joelle Pineau, co-lead of FAIR, is a co-author on a paper building on this work.
jmlr.org/papers/volume2…

The leader of one company's research org uses his power to tarnish this work & acts incredibly defensively, leader of a peer org builds on it. 3\4
Read 4 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!