TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

We read the paper that forced Timnit Gebru out of Google

334 pointsby DarkContinentover 4 years ago

52 comments

jpm_sdover 4 years ago
Here&#x27;s why I have absolutely no sympathy for Google in this situation.<p>They hired Gebru as a professional thorn in their side. &quot;Come up in here and be a pain in the ass! Tell us what we&#x27;re doing wrong!&quot;, they said. &quot;We&#x27;re Enlightened Corporate America, after all!&quot; She is a chess piece in the game of Wokeness Street Cred.<p>She then proceeded to do the job she was hired for, and now they&#x27;re all &quot;Hey lady, Here At Google that&#x27;s not how we do things&quot;.<p>That said, as a manager, I would have &quot;accepted her resignation&quot; and&#x2F;or fired her without hesitation.
评论 #25315124 未加载
评论 #25314735 未加载
评论 #25315176 未加载
评论 #25315077 未加载
评论 #25320213 未加载
评论 #25315518 未加载
评论 #25319239 未加载
评论 #25319024 未加载
评论 #25320640 未加载
评论 #25316737 未加载
评论 #25315580 未加载
评论 #25316506 未加载
评论 #25317593 未加载
评论 #25315287 未加载
sthatipamalaover 4 years ago
Are these numbers the energy to train a model? The whole point of these new NLP models is transfer learning, meaning you train the big model once and fine-tune it for each use case with a lot less training data.<p>5 cars worth of carbon emissions is not a lot given that it is a fixed cost. Very few are retraining BERT from scratch.<p>EDIT:<p>The other two points are also disingenuous.<p>* &quot;[AI models] will also fail to capture the language and the norms of countries and peoples that have less access to the internet and thus a smaller linguistic footprint online. &quot;<p>NLP in &quot;low resource&quot; languages is a major area of research, especially because that&#x27;s where the &quot;next billion users&quot; are for Big Tech. Facebook especially is financially motivated to solve machine translation to&#x2F;from such languages. <a href="https:&#x2F;&#x2F;ai.facebook.com&#x2F;blog&#x2F;recent-advances-in-low-resource-machine-translation&#x2F;" rel="nofollow">https:&#x2F;&#x2F;ai.facebook.com&#x2F;blog&#x2F;recent-advances-in-low-resource...</a><p>* &quot;Not as much effort goes into working on AI models that might achieve understanding, or that achieve good results with smaller, more carefully curated datasets (and thus also use less energy).&quot;<p>This is also a major area of research. Achieving understanding falls under the purview of AGI, which itself carries ethical and safety concerns. There are certainly research groups working toward this. And reducing parameter sizes of big networks like GPT-3 is the next big race. See <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=24704952" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=24704952</a>
评论 #25312909 未加载
评论 #25314595 未加载
评论 #25314576 未加载
评论 #25312853 未加载
评论 #25314728 未加载
hezagover 4 years ago
I did not read the paper (just like most people here), but by the title — “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?” — it does not look like the CO2 emissions thing is the main topic of this research.<p>BTW, &quot;Stochastic Parrots&quot; is a very descriptive name for the problem<p>&gt; Moreover, because the training datasets are so large, it’s hard to audit them to check for these embedded biases. “A methodology that relies on datasets too large to document is therefore inherently risky,” the researchers conclude. “While documentation allows for potential accountability, [...] undocumented training data perpetuates harm without recourse.”<p>Since these models are being applied in a lot of fields that directly affects the life of millions of people, this is a very important and underdiscussed problem.<p>I really want to read the paper.
评论 #25312192 未加载
评论 #25319966 未加载
评论 #25315439 未加载
评论 #25313464 未加载
评论 #25314497 未加载
saeranvover 4 years ago
I feel very disheartened that whatever good Timnit Gebru is doing in her work seems to be undermined by her constant antagonism and hostility. Based on this article, and additional context from other Google employees, I think she is in the right, but that she herself is such an obnoxious person it&#x27;s hard for a lot of us here to be able to separate the issue with her personality.<p>My first introduction to Gebru (and probably a lot of others here) was through her fight with Yann LeCun. Whatever everyone thinks about that debate, one thing was clear: LeCun consistently asked his supporters to not attack Gebru, and tried to engage the discourse, whereas she repeatedly encouraged her supporters to attack LeCun, and avoided his points.<p>FWIW, I myself am a minority. I am very sympathetic to the investigation of AI ethics and Gebru&#x27;s general area of work. I also completely understand and support the need for activists to disrupt norms: we wouldn&#x27;t have civil liberties if activists weren&#x27;t willing to disrupt middle-class norms.<p>But I think for exactly that reason activists have a responsibility (a greater responsibility even) to be selective about which norms or civilities they choose to disrupt. They do such enormous harm when their actions can be picked apart and dismissed for completely unrelated issues - in this case because Timnit doesn&#x27;t seem to engage in good faith discussions with other experts in her field. It sucks, I really hate that this is what ethics in tech is going to be associated with.
评论 #25318452 未加载
eecksover 4 years ago
AI ethics research is funny. It&#x27;s obviously important but it&#x27;s also kind of ... obvious. I am surprised they get paid so much.<p>* Lots of processing uses more energy..<p>* Large amounts of data might contain bad data (garbage in, garbage out)<p>* Wealthy countries&#x2F;communities have the most &#x27;content&#x27; online so less wealthy countries&#x2F;communities will be unrepresented.<p>here&#x27;s some more:<p>* AI being forced to chose between two &quot;bad&quot; scenarios will result in an unfavorable outcome for one party<p>* AI could reveal truths people don&#x27;t want to hear e.g. it might say the best team for a project is an all white male team between 25 - 30 rather than a more diverse team. It might say that a &#x27;perfect&#x27; society needs a homogeneous population.<p>* AI could disrupt a lot of lower paid jobs first without governments having proper supports and retraining structures in place
评论 #25314919 未加载
评论 #25315213 未加载
评论 #25315872 未加载
评论 #25314826 未加载
tinyhouseover 4 years ago
Last comment about it since I spent too much time on this story. As a researcher in the field I feel bad people don&#x27;t give Google more credit (I have no affiliation with Google). They created a research environment where researchers have freedom to work on their own interests and publish papers (they publish more than any other company). You don&#x27;t find many environments like that outside of academia. I still remember Microsoft Research closing down their research lab in the west coast and sending a huge number of researchers home. I can tell you I always apply when there is an opening. So far without luck. If you&#x27;re a Google Researcher don&#x27;t forget how lucky you are.
评论 #25315555 未加载
评论 #25314703 未加载
thinkingemoteover 4 years ago
Anyone else see the narrative on Twitter to be so much different than Reddit and hacker news?<p>On reddit and hacker news I have never seen any unconditional support for Timmit. Even amongst those anti Google and broadly in support of her, there&#x27;s no whole agreement with all of what Timmit says; the assertion of her story, the reasons why and the conclusions.<p>that a. She was fired b. That she was fired because of sexism and racism and c. That all research at Google is pointless and should stop until reform.<p>On Twitter there is more vocal support, but here and Reddit even on those most critical of Google and behind Timmit, there&#x27;s no entire agreement with her. There are comments supporting some parts but none supporting all, none agreeing with the Google Walkout document. Comments pointing out holes with official explanations but not supporting the conclusions.<p>Why is this? I find it odd, I would expect more unequivocal solidarity seen with less public visibility.
评论 #25314832 未加载
评论 #25314531 未加载
评论 #25314746 未加载
评论 #25314462 未加载
评论 #25314557 未加载
评论 #25315902 未加载
评论 #25314938 未加载
评论 #25314721 未加载
评论 #25314480 未加载
评论 #25315006 未加载
dangover 4 years ago
The threads so far on this ongoing story:<p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25307167" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25307167</a><p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25292386" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25292386</a><p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25285502" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25285502</a><p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25289445" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25289445</a>
评论 #25312812 未加载
treisover 4 years ago
This doesn&#x27;t strike me as anything that needs to be buried. The energy argument is tenuous and at this point models use relatively little compute power. The language bias is a better tack but I don&#x27;t think this is earth shattering to anyone. Most of the internet is from Western sources and some fraction of that is racist&#x2F;prejudiced. I think this is obvious to anyone who has ever used the internet.
评论 #25311871 未加载
评论 #25312252 未加载
choegerover 4 years ago
While I will not dive into the extremes here, this <i>is</i> yet another battle in the fight over language, right? The most important point of criticism seems to be that the language models are &quot;unethical&quot;, i.e., not subject to control by a particular political standard.<p>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 <i>proactively</i> controlling you? Should it not always be the <i>output</i> that will be judged <i>by the public</i>?
评论 #25314638 未加载
评论 #25314585 未加载
评论 #25314466 未加载
sumnuyungiover 4 years ago
Is this paper on arxiv? This overview doesn&#x27;t answer any critical questions. For example, it&#x27;s easy to fill up 128 references and a reader shouldn&#x27;t blindly trust a claim that, &quot;The version of the paper we saw does also nod to several research efforts on reducing the size and computational costs of large language models, and on measuring the embedded bias of models.&quot;<p>If a key part of Google&#x27;s claim is that the paper omits relevant research, an author should have simply posted their 128 references and openly asked what work was missing. This whole saga could be easily solved instead of being dragged out for clicks.
评论 #25312251 未加载
评论 #25312007 未加载
KingOfCodersover 4 years ago
See the other thread were one mentions that with reCaptcha Google learns that all cabs are yellow (and traffic lights hang over streets - they don&#x27;t here). But I guess everyone only sees the blind spots of other people.
评论 #25313076 未加载
nullcover 4 years ago
&gt; A version of Google’s language model, BERT, which underpins the company’s search engine, produced 1,438 pounds of CO2 equivalent in Strubell’s estimate<p>So... they&#x27;re saying it used about $100 worth of electricity.<p>[ <a href="https:&#x2F;&#x2F;www.eia.gov&#x2F;tools&#x2F;faqs&#x2F;faq.php?id=74&amp;t=11" rel="nofollow">https:&#x2F;&#x2F;www.eia.gov&#x2F;tools&#x2F;faqs&#x2F;faq.php?id=74&amp;t=11</a> ]<p>[ <a href="https:&#x2F;&#x2F;www.statista.com&#x2F;statistics&#x2F;190680&#x2F;us-industrial-consumer-price-estimates-for-retail-electricity-since-1970&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.statista.com&#x2F;statistics&#x2F;190680&#x2F;us-industrial-con...</a> ]
评论 #25311962 未加载
评论 #25311988 未加载
评论 #25311606 未加载
评论 #25311672 未加载
评论 #25316408 未加载
chairmanwow1over 4 years ago
It&#x27;s nice to finally get to see some of the content of the paper and it&#x27;s awesome that Bender was willing to step up and give some context to world about what the hell it was about.
ethical-aiover 4 years ago
&gt; Though Bender asked us not to publish the paper itself because the authors didn’t want such an early draft circulating online<p>Should not you avoid submitting early drafts which are not good enough to be published yet?
评论 #25314120 未加载
0-_-0over 4 years ago
&quot;...is known for coauthoring a groundbreaking paper that showed facial recognition to be less accurate at identifying women and people of color, which means its use can end up discriminating against them&quot;<p>Wait, isn&#x27;t that the other way around? If it can&#x27;t recognize people of some category then it <i>can&#x27;t</i> be used to discriminate against them, e.g. the police can&#x27;t use it to identify peaceful protesters with those characteristics. I wonder what would have happened if the these networks were much better at recognizing women and people of color, would the paper then be about Google designing technology to detect minorities?
评论 #25313150 未加载
评论 #25314145 未加载
评论 #25313122 未加载
sokoloffover 4 years ago
&gt; An AI model trained on vast swaths of the internet won’t be attuned to the nuances of this vocabulary and won’t produce or interpret language in line with these new cultural norms.<p>As those become pervasive cultural norms, why would the model not adapt to include them?
评论 #25321160 未加载
rayinerover 4 years ago
&gt; An AI model taught to view racist language as normal is obviously bad. The researchers, though, point out a couple of more subtle problems. One is that shifts in language play an important role in social change; the MeToo and Black Lives Matter movements, for example, have tried to establish a new anti-sexist and anti-racist vocabulary. An AI model trained on vast swaths of the internet won’t be attuned to the nuances of this vocabulary and won’t produce or interpret language in line with these <i>new cultural norms.</i><p>This is a pretty superficial take on what is an extremely interesting sociological topic. (To be clear, I’m referring to the article, not the underlying paper which we don’t have.) Obviously just because social movements “have tried to establish ... vocabulary” doesn’t meant that vocabulary has become a “new cultural norm.” Plenty of such efforts end up being cultural dead-ends.<p>Take for example a term like “LatinX.” This term has been proposed and is used by certain people, but is extremely unfamiliar and often alienating to Latinos themselves: <a href="https:&#x2F;&#x2F;www.vox.com&#x2F;2020&#x2F;11&#x2F;5&#x2F;21548677&#x2F;trump-hispanic-vote-latinx" rel="nofollow">https:&#x2F;&#x2F;www.vox.com&#x2F;2020&#x2F;11&#x2F;5&#x2F;21548677&#x2F;trump-hispanic-vote-l...</a> (“[O]nly 3 percent of US Hispanics actually use it themselves.... The message of the term, however, is that the entire grammatical system of the Spanish language is problematic, which in any other context progressives would recognize as an alienating and insensitive message.”).<p>The article hand-waves away a deeply interesting question: What <i>should</i> an AI do here? Should AI reflect society, or be a vehicle for accelerating change? It seems at least reasonable to say that the AI should reflect what people actually say, in which case a big training dataset is appropriate, instead of what some experts decide that people should say. In some contexts, for example with “LatinX,” researchers seeking to enhance inclusivity could instead end up imposing a kind of racist elitism. (People without college educations—which disproportionately comprises immigrants and people of color—tend to be less knowledgeable about and slower to adopt these changes in vocabulary.)<p>The paper seems to imply that AIs should not reflect “social norms” but that training data should be selected to accentuate “attempt[ed]” shifts in such norms. Maybe that’s true, but it doesn’t seem obviously true. To return to the example above, is some Google AI generating the phrase “LatinX” (which 3&#x2F;4 of Latinos have never even heard of: <a href="https:&#x2F;&#x2F;www.pewresearch.org&#x2F;hispanic&#x2F;2020&#x2F;08&#x2F;11&#x2F;about-one-in-four-u-s-hispanics-have-heard-of-latinx-but-just-3-use-it" rel="nofollow">https:&#x2F;&#x2F;www.pewresearch.org&#x2F;hispanic&#x2F;2020&#x2F;08&#x2F;11&#x2F;about-one-in...</a>) in preference to “Latino” or “Hispanic” actually the desired result?
评论 #25313185 未加载
评论 #25315016 未加载
评论 #25312799 未加载
ur-whaleover 4 years ago
It&#x27;s not the paper that forced her out: she forced herself out.
评论 #25328197 未加载
bartreadover 4 years ago
Good grief - this terrible clickbaity headline writing: &quot;We read the paper that forced Timnit Gebru out of Google. Here’s what it says&quot;<p>The paper <i>DID NOT</i> force her out of Google. Her subsequent behaviour - submitting without approval, rant, ultimatum, and resignation - did. And she wasn&#x27;t &quot;forced out&quot;: she resigned of her own volition. She could have chosen to make improvements to the paper based on the feedback she was given, resubmit it for approval, and then get on with her life, but she went the other way.<p>The headline from the last discussion on Timnit&#x27;s exit[0] was awful as well: &quot;The withering email that got an ethical AI researcher fired at Google&quot;. So bad in fact that it was changed on HN to more accurately reflect what actually happened: &quot;AI researcher Timnit Gebru resigns from Google&quot; (much more neutral and factual in tone).<p>Seriously, what happened to journalistic standards and integrity? Why are the actual events being so forcefully twisted to fit a particular narrative? No wonder the general population struggle to figure out what&#x27;s true and what&#x27;s not, and fall victim to all kinds of nonsense, BS theories, and conspiracies.<p>I wish I had a good idea on how to change this behaviour by journalists and publications.<p>(Clearly this is a problem that goes far beyond Timnit&#x27;s story.)<p><i>[0] <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25292386" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25292386</a></i>
评论 #25313994 未加载
评论 #25313275 未加载
评论 #25313280 未加载
评论 #25313403 未加载
评论 #25313770 未加载
评论 #25314110 未加载
评论 #25314995 未加载
评论 #25314408 未加载
_cs2017_over 4 years ago
Despite what both sides claim, IMHO papers like this are primarily PR, ideology, and politics, rather than science and technology.<p>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.
评论 #25312960 未加载
评论 #25313304 未加载
评论 #25312962 未加载
评论 #25313565 未加载
评论 #25312959 未加载
评论 #25312925 未加载
评论 #25313037 未加载
darawkover 4 years ago
These points are...almost uniformly terrible.<p>&gt; Training large AI models consumes a lot of computer processing power, and hence a lot of electricity. Gebru and her coauthors refer to a 2019 paper from Emma Strubell and her collaborators on the carbon emissions and financial costs of large language models. It found that their energy consumption and carbon footprint have been exploding since 2017, as models have been fed more and more data.<p>You train the model once...and then use it to provide incredibly cheap value for billions of people. Comparing the carbon footprint of a single flight between NYC and LA to the training of a model is...insanely disingenuous. The model gets trained once. The correct comparison would be between the carbon footprint of <i>building the plane</i>. Or, atlernatively, to amortize the carbon footprint of the training over all of the individual queries it answers.<p>&gt; Large language models are also trained on exponentially increasing amounts of text. This means researchers have sought to collect all the data they can from the internet, so there&#x27;s a risk that racist, sexist, and otherwise abusive language ends up in the training data.<p>This is the only actually legitimate point. This is a real problem, but everyone already knows about this problem, and so if she&#x27;s going to talk about it she should be doing so in a solutions focused way if she means to contribute anything to the field. She may have done that in the paper, but this review doesn&#x27;t say so.<p>&gt; The researchers summarize the third challenge as the risk of “misdirected research effort.” Though most AI researchers acknowledge that large language models don’t actually understand language and are merely excellent at manipulating it, Big Tech can make money from models that manipulate language more accurately, so it keeps investing in them. “This research effort brings with it an opportunity cost,” Gebru and her colleagues write. Not as much effort goes into working on AI models that might achieve understanding, or that achieve good results with smaller, more carefully curated datasets (and thus also use less energy).<p>Your criticism is that...tech companies are spending capital on making more profit for themselves? Thats uh, not much of a criticism. Especially when you consider the fact that this technology has positive spillover effects for other groups. These language models can be repurposed to combat racism online, and for all sorts of other things. But even if you ignore that, the premise here is just an utterly trivial near-tautology: &quot;Company invests in things that make it more money&quot;.<p>&gt; The final problem with large language models, the researchers say, is that because they’re so good at mimicking real human language, it’s easy to use them to fool people. There have been a few high-profile cases, such as the college student who churned out AI-generated self-help and productivity advice on a blog, which went viral.<p>Sure. You could say this about photoshop, too, and people have. But this technology is going to happen, with or without Google&#x27;s help.<p>&gt; In his internal email, Dean, the Google AI head, said one reason the paper “didn’t meet our bar” was that it “ignored too much relevant research.” Specifically, he said it didn’t mention more recent work on how to make large language models more energy-efficient and mitigate problems of bias.<p>&gt; However, the six collaborators drew on a wide breadth of scholarship. The paper’s citation list, with 128 references, is notably long. “It&#x27;s the sort of work that no individual or even pair of authors can pull off,” Bender said. “It really required this collaboration.”<p>Your defense against a claim that specific research was missed is to cite...the length of the citation list? Lol. This argument would hardly pass muster in a forum comment.
praptakover 4 years ago
A company that profits from AI conducts and oversights research of ethics in AI. What could go wrong?<p>It was just a question of time before something like that blew up.<p>Edit&#x2F;disclaimer: said without judging this particular conflict.
pabl0rgover 4 years ago
Maybe the real problem with the paper is that it points out that GPT-3 and the like can completely distort search results by filling the web with auto-generated spam. Botnets could link spam pages to each other and even generate traffic. If that got into the wrong hands, we would be unable to distinguish truth from fiction.<p>If the general public heard of this, Google’s stock price might be hurt. In such a bot-filled world, humans might prefer to start their searches in walled-gardens or on sites that could better-validate content.
nashaloover 4 years ago
Not totally on topic, but on threads about researchers, I see this trend that when we&#x27;re talking about a male researcher we only mention the family name but when talking about a female researcher we&#x27;re using mostly the first name.<p>Case in point, at time of writing there are 26 mentions of &quot;Timnit&quot; vs 16 mentions of &quot;Gebru&quot; on this thread.<p>I don&#x27;t think there are bad intentions behind this, but it really comes off as infantilizing so maybe we&#x27;d be better off calling her &quot;Gebru&quot;?
评论 #25319533 未加载
评论 #25316342 未加载
评论 #25318588 未加载
评论 #25315937 未加载
mensetmanusmanover 4 years ago
The lifetime output of 5 cars is nothing in a nation of over 300,000,000.<p>If AI can manage peak energy usage efficiently and improve the grid by 1% it will have made a positive impact over 100 fold.
hankchinaskiover 4 years ago
I fail to understand how this is even a story worth covering. Sometime people forget you work FOR someone, who&#x27;s already paying top dollars for your work. If your interests are not aligned with your employer you certainly have the right to walk away. You surely do not have the right to be seen as a beaten up dog when this was exactly what happens when you ignore feedback, rant and give ultimatums like you count more than anybody else. I am glad she was let go
fastballover 4 years ago
If the authors aren&#x27;t confident enough in their paper to release it publicly, why would I want to read <i>someone else&#x27;s</i> (presumably inferior) summary of it?
评论 #25313186 未加载
actuatorover 4 years ago
&gt; It will also fail to capture the language and the norms of countries and peoples that have less access to the internet and thus a smaller linguistic footprint online. The result is that AI-generated language will be homogenized, reflecting the practices of the richest countries and communities.<p>It seems amusing to read this when the same person and her followers hounded Yann LeCun for pointing out the same thing with image models.<p>Anyway, this seems interesting. But I am not sure how you solve this. Do we take representative dataset according to population of a place? Also, assuming this is limited to a single language. How can AI generated language account for nuances from regional differences at the same time in a common model. Isn&#x27;t what the author asking for here is kind of train, en_US, en_GB, en_IN separately here. For things like completion don&#x27;t language models already account for this?
评论 #25312303 未加载
mindrunnerover 4 years ago
Reading Jeff&#x27;s email plus her comments on twitter doesn&#x27;t give the full story.<p>It seems like<p>1. She did not give them the time required to vet through the paper or followed the processes, plus her email to everyone to stop work on other projects. 2. Google fired her immediately, which might have been different if she wasn&#x27;t a POC.<p>This
评论 #25313883 未加载
lsiebertover 4 years ago
Ultimately this reads as straight censorship by Google. Or if not by Google, by Google employees, with the tacit approval of their higher ups.<p>They didn&#x27;t like the viewpoint she expressed, didn&#x27;t like the criticisms she raised, so they blocked her (well her and 7 co-author&#x27;s) paper. When she said that was unacceptable and stood her ground, they badmouthed her and fired her.<p>You don&#x27;t have to agree with the paper&#x27;s criticisms (and it appears they were just part of a longer paper) to be concerned by viewpoint censorship. If the paper wasn&#x27;t worthwhile or based on facts, then that would have come out in academic review, either in peer review in the paper, or in subsequent papers rebutting it, or pointing to subsequent changes. That&#x27;s how academic inquiry works.<p>But if companies can silence ethics researchers who express concerns, whose job, as AI ethicists, is to express concerns, that fundamentally undermines academic inquiry into the topics at hand.
评论 #25313660 未加载
评论 #25313694 未加载
karmasimidaover 4 years ago
Honestly question to fellow HN commenters: what is the popular take on energy efficiency against entertainment industry? Where do people draw the line, regarding what and what not should be an ethnical use case for energy?
godelskiover 4 years ago
While this is a lot of electricity and CO2 to consider, is it fair to compare them to cars? I don&#x27;t think this so much means that one should get rid of training these big models but rather do their best to ensure that the data centers that these models are trained on are getting their electricity from mostly 0 emission sources. This actually seems like a doable task since you don&#x27;t need to be physically near your compute cluster. I think Google and Facebook could also turn this into a PR move by putting some research money into green tech (ambiguous).
ipsocannibalover 4 years ago
I&#x27;m curious if the paper posits any solutions to the risks highlighted? There was mention of references to work on making model training more energy efficient but that was it.
auggieroseover 4 years ago
Doesn&#x27;t make the bar for Google publishing quality standards? That is the most ridiculous thing I have ever heard.
nickdothuttonover 4 years ago
Kicking out (or having resign) an AI researcher is a lot less costly in PR terms than having a large scale AI fail like the one Google experienced when their photo software started identifying black people as monkeys in 2015. Gebru is a fuse that just went pop, and can now be replaced with a new one quite easily.
ChrisArchitectover 4 years ago
What was this article from technology review - yes it gave insight into the contents of the paper but then barely made any conclusion or added anything about what that meant to the situation other than the very end saying the obvious maybe this cuts into Google&#x27;s &#x27;cash cow&#x27;?<p>Bah
juanbyrgeover 4 years ago
I was a little disappointed in the quality of this research. I can see where Google is coming from. Some of these topics seem regurgitated (e.g. bias in online text) and some are just irrelevant (CO2 emissions). Training language models does not contribute in any meaningful way to global CO2 emissions. Overall not a very strong paper in contrast to the other reactions I&#x27;ve been seeing online.
评论 #25312030 未加载
评论 #25312415 未加载
TnkBldrover 4 years ago
what’s the energy expense of training GPT-3? It should be very high. Eager to know the upside of such models. Environmental and social impacts of tech are becoming more direct and mainstream in today’s world!
评论 #25313452 未加载
TnkBldrover 4 years ago
What is the energy expended on training GPT-3? Eager to know the upsides of such models. The environmental and social impacts of tech are more direct and mainstream in today’s world.
评论 #25312334 未加载
WanderPandaover 4 years ago
Thats the kind of problems you get when you let your company turn into a political party...
tinyhouseover 4 years ago
This story blew out of proportion. Social media is sick. Some people&#x27;s tweets read like she is some kind of a George Floyd level victim. She had the best job one can have. Lead researcher who got resources to build a team to work on her own research interest, having all the freedom. And making a killing doing it. She blew it and her race has nothing to do with what happened. She will be fine, joining academia as a tenured professor most likely. If you care about social justice focus on real victims who really need help. Not about this privileged researcher.
mimikatzover 4 years ago
Is this what esteemed PhD AI ethicists spend their time on? It sounds like a useless grad school paper. I think much less of everyone now.
评论 #25312205 未加载
评论 #25312090 未加载
评论 #25312080 未加载
chaganatedover 4 years ago
&gt; There have been a few high-profile cases, such as the college student who churned out AI-generated self-help and productivity advice on a blog, which went viral.<p>This is how much of the internet has felt for a long time. After this nugget, I now wonder if Vox is just a model trained on Piketty and Tumblr.<p>edit: Also not sold on the CO2 argument. Too many variables! Nerds will calculate and re-calculate such things, with the result jumping all over the place, swearing that they&#x27;ve gotten it right--this time! No humility, in spite of the odds.
评论 #25312852 未加载
machinelaboover 4 years ago
I have a genuine question:<p>If a person of color (I am myself non-white) is not performing, what does it take to fire them without the entire world playing the race card on you?<p>Are we creating a society that makes it impossible to fire a person of color? You know there are bad apples in every race, right? How do we handle such scenarios? Seems unfair to me, myself being a person of color - I don&#x27;t want the world to treat me like some kind of a hero for being non-white &#x2F; minority. I want fairness and it is frankly offensive.<p>I am not pleased with the way we&#x27;re treating each other. It&#x27;s supposed to be equal opportunity.<p>I also want us to have scientific discussion about gender differences (backed by research) and other difficult conversations. Nature doesn&#x27;t give a fuck about any of this - if our goal is to uncover the way mother nature works, we&#x27;re going to have to meet difficult truths and not be afraid of it.
评论 #25312253 未加载
评论 #25312088 未加载
评论 #25312287 未加载
评论 #25312157 未加载
评论 #25312217 未加载
评论 #25312215 未加载
评论 #25312260 未加载
评论 #25312206 未加载
评论 #25312132 未加载
评论 #25312552 未加载
评论 #25312599 未加载
评论 #25312363 未加载
评论 #25312093 未加载
mikaelumanover 4 years ago
The problem I find with this has much to do with character. I will not comment on the research (because I do not know the field) but rather comment on why character matters and why motivation is important.<p>Having read her email and tweets; it is quite clear that she is as much a political activist for a far-left “woke” interpretation of the world that I view as of immediate threat to our way of living, democracy and free speech.<p>Of course you can still author great papers but you would be naive to expect her to reach any conclusion that goes against her political goals.<p>In her email she makes this clear. She even posts demands if her paper is not published. In her email she professes to being “dehumanized” for her color and makes it clear that - irrespective of any factors - hiring only 14% women in a year is not acceptable.<p>She continues to say that they had been “enough talking”. We need action.<p>This type of urgency to take action (of course only the actions she herself approves of) and stop talking are clear indications of a person who no longer lives by liberal values but has embraced an ideology to which they now belong.<p>At that point, as an employer of researchers, I would consider her credibility to be severely damaged. If at that point I receive demands from this person “or else”. Then I would of course also accept her resignation.<p>I think the tech industry will need to continue to stand up to the anti-liberal values of these “woke” people before they cause too much harm.
评论 #25313658 未加载
评论 #25314332 未加载
评论 #25313902 未加载
评论 #25314334 未加载
评论 #25314172 未加载
评论 #25313996 未加载
mlthoughts2018over 4 years ago
The summarized sections of the paper seem very, very weak to me.<p>1. Environmental footprint of technology must always be considered as a trade-off for what you get in return. Why do we spend energy on a giant render farm for Pixar? Because the cinematic artwork is worth that environmental cost for many people. Obviously we should pursue improvements in environmental outcomes, but that is not a goal unto itself in a vacuum apart from all other side effects of a technology. Is it worth ~5 car-lifetimes to train GPT? I would say overwhelmingly yes. It reminds me of an anecdote from The Beginning of Infinity by David Deutsch, where some ethicists argued about whether it was a useful human endeavor to invent color TV monitors back when they first hit the market. Why would you need to spend resources creating something besides black and white TV? Yet today and for decades, color monitors are used as critical tools to diagnose diseases and save lives.<p>2. Nobody is required to accept “wokeness” vocabulary, and indeed one of the signs of a crank or a quack is making up a fiefdom of new vocabulary and collecting rent in the form of social authority for the validity and requirement of that new vocab. Nobody is required to be on the cutting edge of how activist language changes, and it seems like a disingenuous way to try to make a cottage industry out of one’s own expertise in rapid changes to activist language. As long researchers are stating what corpus is used, and making tools available that allow peer reviewers to audit that corpus, they are meeting their obligations to their professional field and to society. We should be <i>happy</i> that language researchers would produce lots of papers and lots of models on many sets of corpora, and as the cost of training these models gets cheaper, and the cost of curating the corpora gets cheaper, we can expect to see better variety of curated large corpora, domain-specific corpora and other things.<p>3. Researcher opportunity cost is perhaps the most ridiculous objection. Researchers are free agents to decide what they want to study. In most PhD programs, especially in machine learning, the project selection is entirely up to the student. If Timnit wants there to be different research priorities, well, news flash, but she is only one of eight billion. Unless she wants to raise money to give as research grants that tie the researchers to her pre-decided methods of inquiry, she really has nothing to say here.<p>4. Inscrutable models - this is the only one where there is any point. <i>If</i> the models can produce harmful outcomes and they are inscrutable, then debugging or safeguarding them is a real problem. But this has been true for almost all types of computer science algorithms. Of course we should work on methods that synthesize clarity from the prediction mechanism of large neural nets, but that is also <i>not</i> a criticism of neural nets. That’s just a need for <i>more technology.</i><p>Overall the main points of this paper seem full of themselves, arrogant and overly self-important, especially with wildly ridiculous connections to “wokeness” vocabulary.<p>Given the immediate nuclear option of the ultimatum and email that Timnit sent, I suspect the full text of the paper would be even more unacceptable.<p>Google has plenty of bullshit issues with the way it treats employees and transparency of decision making. Rejecting this publication approval was not one of them.
tsjqover 4 years ago
Wow. The pro-google push is really visible in this thread
评论 #25313908 未加载
评论 #25313717 未加载
alangibsonover 4 years ago
&gt; I understand the concern over Timnit’s resignation from Google - Jeff Dean<p>That&#x27;s just outright false. She was terminated, effective immediately.<p>Does anyone else find themselves losing a lot of respect for Jeff Dean in all this?
评论 #25313126 未加载
评论 #25313053 未加载
评论 #25313107 未加载
vmceptionover 4 years ago
Googlers are waaay too comfortable with their message board system and if they stay there long enough and keep interacting with it they’ll get burned
quantumwokeover 4 years ago
Dr Gebru&#x27;s research AFAICT shines a valuable light on some interesting ethical problems. Yet the response so far from Google and even HN has been one of censorship and suppression of Dr Gebru&#x27;s free speech. How can we reconcile free speech and Google&#x27;s right of review to her research?
评论 #25312186 未加载
ioctrlover 4 years ago
As a white man, when my work is rejected, I assume it&#x27;s because my work needs improvement. This is an advantage of being a white man. When things don&#x27;t go my way, I don&#x27;t need to wonder if it&#x27;s a macro-aggression, an act of systemic discrimination, or suppression of a marginalized voice. I don&#x27;t envy my colleagues who navigate these waters.<p>Furthermore, even if my manager or employer makes it an explicit priority to promote more women or PoC, I don&#x27;t question whether my own gender or skin color will ultimately work against me at promotion time. I grew up in a time and environment where I didn&#x27;t need to question that, so I tend not think about it. Even if I did think about it, it&#x27;s a culturally inappropriate question to ask, so my only option is to keep my head down and work harder.
评论 #25317527 未加载