5 ms·

I think one issue here is that Twitter makes it hard to have good conversations because it motivates people to optimize for pithiness and likeability/retweetabi

I think one issue here is that Twitter makes it hard to have good conversations because it motivates people to optimize for pithiness and likeability/retweetability over substance. That said, I'm disappointed at the lack of professionalism shown by Timnit Gebru in this conversation. I understand that she's frustrated that mainstream AI/ML research doesn't have a good understanding of how models can reflect and reinforce systemic racism. But her response to Yann doesn't attempt to find common ground, doesn't add any meaningful facts to the conversation, and is ad-hominem ("why are YOU too dumb to understand this?"). Timnit is a Research Scientist at Google who leads a group that works on Ethical AI. In that role I'd expect that her mission would be to elevate the collective understanding of ethical problems in AI, and solutions to those problems. Her communication here doesn't achieve that. This is important to me because I'm also a (fledgling) member of the AI/ML community, and I want us to be able to have civil and constructive conversations.

excerionsforte 6y ago
"why are YOU too dumb to understand this?" Citation needed. I don't like this sensationalism. "civil and constructive conversations." What was uncivil nor constructive? One person blamed a dataset for bias, but said researchers should not be on the hook for consequences stemming from the output that could be put into a deployed product. The other disagreed with the notion and stating that this was brought up in the past showing clear frustration. Uncivil tends to be a dogwhistle indicating how a person SHOULD act as if there are strict set of parameters for how to show frustration.
- PixyMisa 6y ago
  So someone is frustrated because they persist in blaming the wrong people for a problem? Whose fault is that?
  - excerionsforte 6y ago
    Awareness and requests for understanding is not equal to blame. Please cite where you see the blaming.
- kortilla 6y ago
  > Uncivil tends to be a dogwhistle People who cry dogwhistle instead of addressing the point tend to be more interested in shaming people for accidentally saying something undesirables say than engaging in constructive discourse.
  - excerionsforte 6y ago
    Who cried dogwhistle? You can perceive what you want, but that does not make it true. I'm not going to go back and forward on this kind of pedantry. I asked a question, requested citations for alleged perceptions of incivility and will only respond to answers to it.
    - kortilla 6y ago
      > Who cried dogwhistle? You. There is no point in telling people that there are dogwhistles in their argument. The extra woke points it gives you don’t help with people interested in logical discussions. > You can perceive what you want, but that does not make it true. I’m not perceiving anything. It’s a term you used right in your comment. I’m just telling you that it’s a term used to divide and it does nothing to support a broader point.
- visarga 6y ago
  Timnit Gebru quotes: > Yann, I suggest you watch me and Emily’s tutorial or a number of scholars who are experts in this. In other words, poor Yann should be sent for reeducation! > Maybe your colleagues will try to educate you. Maybe not. But I have better things to do than this. Or maybe not. So much thinly veiled despise. If it weren't for Yann there would be no computer vision as it is today, and her job would be the realm of sci-fi. > If there is any sort of fairness, ethics or what have you initiative @Facebook and @facebookai, let me tell you how little credibility they have because you have the loudest microphone out there. Now switching on an indirect strategy - throw shit at his employer in order to force his hand. > I’m sick of this framing. Tired of it. Many people have tried to explain, many scholars. Listen to us. You can’t just reduce harms caused by ML to dataset bias. Yet I have hardly seen a good counterexample of algorithmic bias, it was almost always the data. Gradient descent doesn't know about race. CNNs, RNNs, Transformers - they are all orthogonal to social bias. Maybe regularisation can have an effect on accuracy for small (less represented) classes, but not much more.
  - excerionsforte 6y ago
    """ Known for her work on racial and gender bias in facial recognition systems and other AI algorithms, Gebru has been advocating for fairness and ethics in AI for years. The Gender Shades project that she leads with MIT Media Lab Computer Scientist Joy Buolamwini revealed that commercial facial recognition software was more likely to misclassified and was less accurate with darker-skinned females compared to lighter-skinned men. Gebru’s CVPR 2020 talk Computer vision in practice: who is benefiting and who is being harmed? again addressed the role of bias in AI, “I think that now a lot of people have understood that we need to have more diverse datasets, but unfortunately I felt like that’s kind of where the understanding has stopped. It’s like ‘let’s diversify our datasets. And that’s kind of ethics and fairness, right?’ But you can’t ignore social and structural problems.“ """ Seems that these two paragraphs alone describe what she has been saying. Where did your "algorithmic bias" assertion came from? Who mentioned algorithmic problems? > If there is any sort of fairness, ethics or what have you initiative @Facebook and @facebookai, let me tell you how little credibility they have because you have the loudest microphone out there. Its interesting how one perceives this as "throwing shit" when it is harsh criticism directed at what she denotes as the little power any ethics or fairness org at FB has. The other quotes are out of frustration and are in the civil territory. If those are uncivil to you and the others, well, that's your line of sensitivity as it is definitely not the bar for incivility. If something sounds harsh, it doesn't mean it is uncivil. It is very easy to twist words or remove them out of the conversation context to make them seem like fighting words. I could be very harsh in address people here and still be civil about it. The air I smell here 'It is Yann LeCunn, legendary figure, never criticize what he says or if you do so, do it very very very politely even if you have to do it 1k times.' I'm no longer interested in finding out more perceptions on this. I saw what I saw and you guys see what you see. I don't see any uncivil activity just sensationalism at an opportune time and a mob of anti-social justice trying to attack whatever seems "woke." These issues would have to be dealt with no matter what period of time.
- jml7c5 6y ago
  You are correct about the 'quote', as it does not seem Gebru ever wrote that. I am surprised none of these sibling comments have affirmed that. While personally I presume that parent poster meant the 'quote' only as an illustrative pharaphrase, it is clearly very easy to misinterpret. (Of course, if one assigns less charitable intent it is intentionally misleading, but I generally assume the best of people.) Either way, it feels intellectually dishonest that no one else has conceded this. As an aside: I wish English had some more elegant method for forming compound words than just-concatenate-words-with-hyphens. Compounding of phrases and use of imagined quotes as compound words is one of those common flourishes of verbal communication that is difficult to express in written English. As a note: I believe it is primarily the "dogwhistle" reference is what is earning you down-votes, as it subtly implies — whether intended or not — that parent poster is somehow racist. I presume the best intentions and up-voted you, but keep in mind how it can be interpreted.
  - excerionsforte 6y ago
    I'm glad there are a few people who looked at what I said objectively. I only asked for citations and noted my reason for asking. Had not seen one citation but sensationalism and perceived attitude from text. I've had people call my behavior "unprofessional" in the past because they had a very narrow set of expectations for "professional." Whatever that seemed to fit them per person. Always glad to have a third party to sort it out with them. I regard unprofessional as a potential dogwhistle these days since the only people who understand what it really means are the ones who knows it is being used in a narrow subjective manner. They are just loaded terms.
  - KerrickStaley 6y ago
    "why are YOU too dumb to understand this?" is not a quote and was not meant to be; it's a paraphrase/interpretation of what she wrote. I think this is fairly obvious in the context of my comment.
PixyMisa 6y ago
Twitter is designed quite effectively to prevent constructive conversations.
- randomsearch 6y ago
  I think it’s also the way people use it. Are you interested in truth or power? It seems Twitter attracts a lot of people who only feel powerful on twitter, and they prioritise that over truth.
- max_ 6y ago
  How would one design a social network focused on promoting constructive conversations.
gedy 6y ago
> In that role I'd expect that her mission would be to elevate the collective understanding of ethical problems in AI, and solutions to those problems To butcher the phrase a bit: “It is difficult to get a wo/man to correct something, when his/her salary depends on not fixing it.”
- tinyhouse 6y ago
  Fixing something in AI usually means improving it, not solving it. And even if the problem would get solved, there are plenty of other problems to work on.
  - thu2111 6y ago
    Maybe not for her. Seems her education is actually electrical engineering for some reason, and her Wikipedia page lists no record of AI research achievement beyond racial/sociological stuff. If she accepted that the USA/AI isn't filled with racism she'd presumably have to go back to signal processing.
    - r-zip 6y ago
      FYI: education in EE is actually pretty common for people in ML. The math needed for control and signal processing has a lot of overlap with ML, and many EE labs do research in ML. I don't agree with her responses to LeCun, but I don't think her education should impact her credibility. Edit: If you read her bio, she received her PhD in CS, and it looks like she's published multiple papers on things other than social issues in ML. What are you even talking about?
      - thu2111 6y ago
        I was going based on her Wikipedia page which doesn't seem to mention much beyond racial stuff.
zucker42 6y ago
> I think one issue here is that Twitter makes it hard to have good conversations because it motivates people to optimize for pithiness and likeability/retweetability over substance. Which is why it's probably a good thing on balance that Yann LeCun left Twitter. Blog posts and comments are the bare minimum for an adequate forum for discussions like this, since the reader is at least encouraged to read and understand an accurate and complete summary of the author's points before making a pithy reply[1]. [1] My phrasing does not necessarily refer to any of the figures involved in this conversation.

excerionsforte 6y ago

"why are YOU too dumb to understand this?" Citation needed. I don't like this sensationalism. "civil and constructive conversations." What was uncivil nor constructive? One person blamed a dataset for bias, but said researchers should not be on the hook for consequences stemming from the output that could be put into a deployed product. The other disagreed with the notion and stating that this was brought up in the past showing clear frustration. Uncivil tends to be a dogwhistle indicating how a person SHOULD act as if there are strict set of parameters for how to show frustration.

PixyMisa 6y ago
So someone is frustrated because they persist in blaming the wrong people for a problem? Whose fault is that?
- excerionsforte 6y ago
  Awareness and requests for understanding is not equal to blame. Please cite where you see the blaming.
kortilla 6y ago
> Uncivil tends to be a dogwhistle People who cry dogwhistle instead of addressing the point tend to be more interested in shaming people for accidentally saying something undesirables say than engaging in constructive discourse.
- excerionsforte 6y ago
  Who cried dogwhistle? You can perceive what you want, but that does not make it true. I'm not going to go back and forward on this kind of pedantry. I asked a question, requested citations for alleged perceptions of incivility and will only respond to answers to it.
  - kortilla 6y ago
    > Who cried dogwhistle? You. There is no point in telling people that there are dogwhistles in their argument. The extra woke points it gives you don’t help with people interested in logical discussions. > You can perceive what you want, but that does not make it true. I’m not perceiving anything. It’s a term you used right in your comment. I’m just telling you that it’s a term used to divide and it does nothing to support a broader point.
visarga 6y ago
Timnit Gebru quotes: > Yann, I suggest you watch me and Emily’s tutorial or a number of scholars who are experts in this. In other words, poor Yann should be sent for reeducation! > Maybe your colleagues will try to educate you. Maybe not. But I have better things to do than this. Or maybe not. So much thinly veiled despise. If it weren't for Yann there would be no computer vision as it is today, and her job would be the realm of sci-fi. > If there is any sort of fairness, ethics or what have you initiative @Facebook and @facebookai, let me tell you how little credibility they have because you have the loudest microphone out there. Now switching on an indirect strategy - throw shit at his employer in order to force his hand. > I’m sick of this framing. Tired of it. Many people have tried to explain, many scholars. Listen to us. You can’t just reduce harms caused by ML to dataset bias. Yet I have hardly seen a good counterexample of algorithmic bias, it was almost always the data. Gradient descent doesn't know about race. CNNs, RNNs, Transformers - they are all orthogonal to social bias. Maybe regularisation can have an effect on accuracy for small (less represented) classes, but not much more.
- excerionsforte 6y ago
  """ Known for her work on racial and gender bias in facial recognition systems and other AI algorithms, Gebru has been advocating for fairness and ethics in AI for years. The Gender Shades project that she leads with MIT Media Lab Computer Scientist Joy Buolamwini revealed that commercial facial recognition software was more likely to misclassified and was less accurate with darker-skinned females compared to lighter-skinned men. Gebru’s CVPR 2020 talk Computer vision in practice: who is benefiting and who is being harmed? again addressed the role of bias in AI, “I think that now a lot of people have understood that we need to have more diverse datasets, but unfortunately I felt like that’s kind of where the understanding has stopped. It’s like ‘let’s diversify our datasets. And that’s kind of ethics and fairness, right?’ But you can’t ignore social and structural problems.“ """ Seems that these two paragraphs alone describe what she has been saying. Where did your "algorithmic bias" assertion came from? Who mentioned algorithmic problems? > If there is any sort of fairness, ethics or what have you initiative @Facebook and @facebookai, let me tell you how little credibility they have because you have the loudest microphone out there. Its interesting how one perceives this as "throwing shit" when it is harsh criticism directed at what she denotes as the little power any ethics or fairness org at FB has. The other quotes are out of frustration and are in the civil territory. If those are uncivil to you and the others, well, that's your line of sensitivity as it is definitely not the bar for incivility. If something sounds harsh, it doesn't mean it is uncivil. It is very easy to twist words or remove them out of the conversation context to make them seem like fighting words. I could be very harsh in address people here and still be civil about it. The air I smell here 'It is Yann LeCunn, legendary figure, never criticize what he says or if you do so, do it very very very politely even if you have to do it 1k times.' I'm no longer interested in finding out more perceptions on this. I saw what I saw and you guys see what you see. I don't see any uncivil activity just sensationalism at an opportune time and a mob of anti-social justice trying to attack whatever seems "woke." These issues would have to be dealt with no matter what period of time.
jml7c5 6y ago
You are correct about the 'quote', as it does not seem Gebru ever wrote that. I am surprised none of these sibling comments have affirmed that. While personally I presume that parent poster meant the 'quote' only as an illustrative pharaphrase, it is clearly very easy to misinterpret. (Of course, if one assigns less charitable intent it is intentionally misleading, but I generally assume the best of people.) Either way, it feels intellectually dishonest that no one else has conceded this. As an aside: I wish English had some more elegant method for forming compound words than just-concatenate-words-with-hyphens. Compounding of phrases and use of imagined quotes as compound words is one of those common flourishes of verbal communication that is difficult to express in written English. As a note: I believe it is primarily the "dogwhistle" reference is what is earning you down-votes, as it subtly implies — whether intended or not — that parent poster is somehow racist. I presume the best intentions and up-voted you, but keep in mind how it can be interpreted.
- excerionsforte 6y ago
  I'm glad there are a few people who looked at what I said objectively. I only asked for citations and noted my reason for asking. Had not seen one citation but sensationalism and perceived attitude from text. I've had people call my behavior "unprofessional" in the past because they had a very narrow set of expectations for "professional." Whatever that seemed to fit them per person. Always glad to have a third party to sort it out with them. I regard unprofessional as a potential dogwhistle these days since the only people who understand what it really means are the ones who knows it is being used in a narrow subjective manner. They are just loaded terms.
- KerrickStaley 6y ago
  "why are YOU too dumb to understand this?" is not a quote and was not meant to be; it's a paraphrase/interpretation of what she wrote. I think this is fairly obvious in the context of my comment.

PixyMisa 6y ago

Twitter is designed quite effectively to prevent constructive conversations.

randomsearch 6y ago
I think it’s also the way people use it. Are you interested in truth or power? It seems Twitter attracts a lot of people who only feel powerful on twitter, and they prioritise that over truth.
max_ 6y ago
How would one design a social network focused on promoting constructive conversations.

gedy 6y ago

> In that role I'd expect that her mission would be to elevate the collective understanding of ethical problems in AI, and solutions to those problems To butcher the phrase a bit: “It is difficult to get a wo/man to correct something, when his/her salary depends on not fixing it.”

tinyhouse 6y ago
Fixing something in AI usually means improving it, not solving it. And even if the problem would get solved, there are plenty of other problems to work on.
- thu2111 6y ago
  Maybe not for her. Seems her education is actually electrical engineering for some reason, and her Wikipedia page lists no record of AI research achievement beyond racial/sociological stuff. If she accepted that the USA/AI isn't filled with racism she'd presumably have to go back to signal processing.
  - r-zip 6y ago
    FYI: education in EE is actually pretty common for people in ML. The math needed for control and signal processing has a lot of overlap with ML, and many EE labs do research in ML. I don't agree with her responses to LeCun, but I don't think her education should impact her credibility. Edit: If you read her bio, she received her PhD in CS, and it looks like she's published multiple papers on things other than social issues in ML. What are you even talking about?
    - thu2111 6y ago
      I was going based on her Wikipedia page which doesn't seem to mention much beyond racial stuff.

zucker42 6y ago

> I think one issue here is that Twitter makes it hard to have good conversations because it motivates people to optimize for pithiness and likeability/retweetability over substance. Which is why it's probably a good thing on balance that Yann LeCun left Twitter. Blog posts and comments are the bare minimum for an adequate forum for discussions like this, since the reader is at least encouraged to read and understand an accurate and complete summary of the author's points before making a pithy reply[1]. [1] My phrasing does not necessarily refer to any of the figures involved in this conversation.