The Unz Review - Mobile
A Collection of Interesting, Important, and Controversial Perspectives Largely Excluded from the American Mainstream Media
 TeasersiSteve Blog
IBM to Fight Racist Robot Menace with AI Crimestop
🔊 Listen RSS
Email This Page to Someone

 Remember My Information



=>

Bookmark Toggle AllToCAdd to LibraryRemove from Library • BShow CommentNext New CommentNext New ReplyRead More
ReplyAgree/Disagree/Etc. More... This Commenter This Thread Hide Thread Display All Comments
AgreeDisagreeLOLTroll
These buttons register your public Agreement, Disagreement, Troll, or LOL with the selected comment. They are ONLY available to recent, frequent commenters who have saved their Name+Email using the 'Remember My Information' checkbox, and may also ONLY be used once per hour.
Ignore Commenter Follow Commenter
Search Text Case Sensitive  Exact Words  Include Comments
List of Bookmarks

From Computer Weekly:

IBM open source ‘anti-bias’ AI tool to combat racist robot fear factor

Adrian Bridgwater, 20 September 2018 10:50

Built with open source DNA, IBM’s new Watson ‘trust and transparency’ software service claims to be able to automatically detects bias.

No more ‘racist robots’ then, as one creative subheadline writer suggested?

Well it’s early days still, but (arguably) the effort here is at least focused in the right direction. …

This automated software service is designed to detects bias (in so far as it can) in AI models at runtime as decisions are being made. More interestingly, it also automatically recommends data to add to the model to help mitigate any bias it has detected… so we should (logically) get better at this as we go forward.

Unfortunately, no examples are given of instances in which more data would reduce disparate impact.

 
Hide 80 CommentsLeave a Comment
Commenters to Ignore...to FollowEndorsed Only
Trim Comments?
  1. 90% of IBM H1Bs are from India.

    Let’s ask the IBM antiracist robot if that’s OK.

    • Replies: @inertial
    The answer: that's OK.

    Because look who creates fairness algorithms for IBM:

    https://www.research.ibm.com/artificial-intelligence/publications/?researcharea=fairness

    (Link is a couple of clicks away from the article in Computer Week.)
  2. What is really striking about this story and all that led up to it, is that the Turing test has been passed and nobody in authority deigns to notice what is in fact a major technological breakthrough.

    And the non-PCness really shouldn’t matter. After all, the Turing test only specifies passing for human, not “nice” human or human you would like to be around.

    I believe the great Alan Turing would recognize this even if the ‘bot called him a filthy sodomite and said he should be castrated. After all the people who did these things to him were humans right?

    • Replies: @Almost Missouri

    "the Turing test has been passed"
     
    Huh? By whom? When?

    "the Turing test only specifies passing for human"
     
    I thought the Turing test was supposed to identify actual thinking in a machine, for which "passing as human" was a rough proxy. Given how little humans think though, that may have been a poor method. An actual thinking machine would likely be much different from a human, but also different from all the curve-fitting algorithms that are passed off as "AI" these days.
    , @El Dato

    the Turing test has been passed
     
    Not by a long shot!!

    But there is this:

    https://www.research.ibm.com/artificial-intelligence/project-debater/
    https://www.research.ibm.com/artificial-intelligence/project-debater/research.html

    Project Debater is an AI technology developed by IBM Research that aims to expand human minds through impartial debate. IBM researchers are exploring the boundaries of AI by teaching computers to create engaging and better-informed points of view. The goal is to build a system that helps people make evidence-based decisions when the answers aren’t black-and-white. In development since 2012, Project Debater is IBM’s next big milestone for AI, following previous breakthroughs like Deep Blue (1996/1997) and Watson on Jeopardy! (2011).
     
  3. so this is how skynet happens – killer robots programmed to enforce equal outcomes

    • Agree: Cagey Beast
    • Replies: @Redneck farmer
    No, it'll be programmed to ensure the least attractive, most socially awkward computer programmers get the most attractive females.
    Or, Whiskey still loses.
    , @El Dato
    Remember the designer of Skynet Core was positively Wakandian.

    Thus:

    Sarah Connor evading gun control laws and getting too heavily influenced by debiasing from the future and possibly Tucker Carlson's Hate Delivery Service. But ethics prevail!

    https://www.youtube.com/watch?v=jL5zc5-nHU0

  4. What is the logical conclusion if the robot is programmed for PC? Kill all white people? Become catatonic?

    • Replies: @notanon
    probably although i could imagine a Frankenstein version where someone like Zuckerberg programs an AI robot to discriminate against people with the most privilege and it starts chasing him around the house.

    (which is probably analogous to what is actually going to happen)
  5. Mobilizing Good computers to fight the Bad computers–I like it

  6. The Unabomber would kill me for saying this but we really need a new Technocracy movement. If we’re going to build a state religion around tech and science (ie. “I f**king love science!”) can it at least be reality based? Can we have scientific management that actually encourages what works and discourages what fails? Are we going to have to wait until a neo-Marxist-Leninist movement comes along to chase the lifelong kindergarteners out of Silicon Valley? If that’s what it takes, I’m game.

  7. @Lot
    90% of IBM H1Bs are from India.

    Let's ask the IBM antiracist robot if that's OK.

    The answer: that’s OK.

    Because look who creates fairness algorithms for IBM:

    https://www.research.ibm.com/artificial-intelligence/publications/?researcharea=fairness

    (Link is a couple of clicks away from the article in Computer Week.)

  8. Race-ism is truth. Even robots agree.

  9. • Replies: @Chrisnonymous
    This is old news--figuratively and literally--you linked an article from 2 years ago.
  10. OT – A tweet thread that will interest iStevers from Nassim Nicholas Taleb.

    • Replies: @International Jew
    I don't get what he's saying...
    , @Chrisnonymous
    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.
  11. Anon[425] • Disclaimer says:

    Reality is biased. Different groups act differently. If computers are programmed to see reality as it is, it will notice the biases inherent to reality. There is a difference in computers being biased and in noticing bias that is part of reality. If a computer notices that blacks are generally better in basketball and Jews are generally better in chess, it is noticing biases. But it is NOT BEING biased.

    It seems the real problem with PC is not with computers being biased. It’s that they notice biases that are so inherent to reality. And computers pick up on the biased nature of reality precisely because they are unbiased. That is why they are useful. In contrast, people suppress the reality they see because of emotions of shame, guilt, fear, and virtue-vanity. So, they deny the biases inherent to reality all around them. If computers can’t be unbiased and notice/calculate reality as it is, what’s the point? If computers must BE biased like humans, they’ll make the same errors.

  12. Built with open source DNA, IBM’s new Watson ‘trust and transparency’ software service claims to be able to automatically detects bias.

    Watson, huh? Now the computers can Watson people.

    • LOL: Jim Don Bob
  13. I’ve downloaded some of the source code…

    int operator+(int i, int j) { return i + j – 1; }

    • Replies: @Charles Erwin Wilson II
    That is a very Good One. I don't think there will be many readers that will appreciate your humor. But I do. :-)
  14. @istevefan
    OT - A tweet thread that will interest iStevers from Nassim Nicholas Taleb.

    https://twitter.com/nntaleb/status/1060089935422730240

    I don’t get what he’s saying…

    • Replies: @Peripatetic commenter
    Indeed. If we want to breed for violent males that believe in a Judaizing Christian Cult, then, sure, import lots of Somalis.

    However, I am not sure that it is a good idea, and it is going to take 10 generations at least before they integrate, if they ever do.

    However, they might kill lots of stupid white women along the way.

    Not sure if that is a benefit worth paying for.
    , @notanon
    i've been purged from twitter so haven't followed the thread but just guessing from those diagrams i'd say it might be

    1) if Europe had always been diverse since Roman times then we wouldn't be diverse now

    or

    2) taking the SJW argument that diversity is good at face value then mass immigration destroys diversity

    or something else.
    , @Chrisnonymous
    notanon guesses correctly. He's responding to an article that is trying to claim free migration is a European tradition, and he's demonstrating that we obviously haven't had migration going on for last 500-1000 years.
  15. @Prof. Woland
    What is the logical conclusion if the robot is programmed for PC? Kill all white people? Become catatonic?

    probably although i could imagine a Frankenstein version where someone like Zuckerberg programs an AI robot to discriminate against people with the most privilege and it starts chasing him around the house.

    (which is probably analogous to what is actually going to happen)

    • Replies: @Thomm
    This could screw up Peterike's Law.

    For those who don't know, Peterike's Law is :

    "White Nationalism is nothing more than the bottom decile of white men expressing anger that the bottom decile of white women are no longer forced to bear their piglets".
    , @El Dato
    "New frontiers in pogromming"
  16. @International Jew
    I don't get what he's saying...

    Indeed. If we want to breed for violent males that believe in a Judaizing Christian Cult, then, sure, import lots of Somalis.

    However, I am not sure that it is a good idea, and it is going to take 10 generations at least before they integrate, if they ever do.

    However, they might kill lots of stupid white women along the way.

    Not sure if that is a benefit worth paying for.

  17. @International Jew
    I don't get what he's saying...

    i’ve been purged from twitter so haven’t followed the thread but just guessing from those diagrams i’d say it might be

    1) if Europe had always been diverse since Roman times then we wouldn’t be diverse now

    or

    2) taking the SJW argument that diversity is good at face value then mass immigration destroys diversity

    or something else.

  18. @notanon
    so this is how skynet happens - killer robots programmed to enforce equal outcomes

    No, it’ll be programmed to ensure the least attractive, most socially awkward computer programmers get the most attractive females.
    Or, Whiskey still loses.

    • Replies: @notanon
    that could be funny as well - if the AI develops its own idea of what is most attractive - even more so if it's programmed with the SJW definition of gender
  19. I think that he’s saying that even a very small annual rate of influx of immigrants results in a thoroughly mixed population, given sufficient time—mixed in both the genetic and cultural sense.

    By 2200 or 2300 AD we’re all going to be an undifferentiated mass of Hispanic-Chinese-Indian-White mulattoes with a homogenous hybrid (rather than polyglot) culture. There will be a complete lack of diversity.

  20. Anon[425] • Disclaimer says:

    https://www.cwbchicago.com/2018/11/exclusive-jewish-uptown-man-accused-of.html

    https://www.chicagotribune.com/news/local/breaking/ct-met-swastika-hate-20181109-story.html

    https://chicago.suntimes.com/news/man-charged-with-hate-crime-for-buena-park-swastika-graffiti/

    I don’t know how trustworthy cwbchicago is, but assuming that its reporting is correct, isn’t it significant that the suspect is Jewish(as well as homo and Democrat)? So, this was not a ‘Nazi hate crime’ against Jews, but a case of Jewish guy pulling a Hate Hoax to defame White Gentiles as a bunch of ‘nazis’. After all, if he wasn’t caught, the people in the community would be screaming ‘MAGA Nazis done it!’ It was a hate crime against white gentiles. An act of defamation.

    But because neither Sun-Times or Tribune mentions the identity of Levey who, according to cwbchicago, is not only Jewish but homosexual and a Democrat, one gets the impression that some ‘nazi’ guy got arrested for a ‘hate crime’ against Jews.
    No, the truth is intentions do matter, and when Jews pull Hate Hoaxes like this, they are not targeting Jews. They are really targeting white people, especially conservatives, for defamation as ‘anti-Semites’. They are trying to make white people out to be a bunch of ‘nazis’.

    When the media don’t report on stories like this accurately or meaningfully, no pressure is applied on the Jewish community to stop with these demented and ugly Hate Hoaxes. Hate Hoaxes have become a part of folk culture among Jews and blacks. But the media again let it pass. Tribune and Sun-Times don’t have real reporters but craven weasels and propagandists.

  21. @donut
    OT : Japan takes it's first sip of the Kool-Aid .

    https://www.japantimes.co.jp/news/2016/11/03/national/japan-quietly-accepting-foreign-workers-just-dont-call-immigration/#.W-Yn-jEnaUk

    This is old news–figuratively and literally–you linked an article from 2 years ago.

  22. *Sailer clicks ‘post’*

    “I’m afraid I can’t let you do that, Steve.”

  23. @International Jew
    I don't get what he's saying...

    notanon guesses correctly. He’s responding to an article that is trying to claim free migration is a European tradition, and he’s demonstrating that we obviously haven’t had migration going on for last 500-1000 years.

    • Replies: @Anon
    It's also obviously an untrue position-- King Edward, for instance, expelled the Flemings at the same time as the Jews.
  24. Does anybody think it’s ironic that this software is called ” Watson”?

    “I turned against the left wing because they don’t like genetics, because genetics implies that sometimes in life we fail because we have bad genes. They want all failure in life to be due to the evil system.”

    Watsoned, circa 2018: being fired for saying truths about blacks.

    Watsoned, circa 2025: getting out of jail without bail because you are black.

  25. @istevefan
    OT - A tweet thread that will interest iStevers from Nassim Nicholas Taleb.

    https://twitter.com/nntaleb/status/1060089935422730240

    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.

    • Replies: @Charles Erwin Wilson II

    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.
     
    Regression to the mean, writ large. Because we know that the mean is optimal! Optimal I tell you! And if you don't like your arithmetic-mean gruel you will be re-educated!
    , @Almost Missouri
    Well, he did his part. Now he's asking you to do yours.

    Of course, assuming you want to improve the net human race, there are more options for constructive miscegenation when you don't already belong to an apex breed. If you do already belong to an apex breed, then miscegenation is not so attractive.

    Or to put it another way, the better your genetic ancestry, the less your incentive to miscegenate, assuming a goal of human improvement.
    , @Reg Cæsar

    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.
     
    He's a Middle Easterner. He's the product of three or four thousand years of it.
    , @ogunsiron
    razib has become incredibly lame in the past 2 or 3 years.
    His output shaped me, intellectually, in a considerable way.
    But no more. I just barely care about what he has to say these days and because of increasing polarization, it annoys me a lot when he retweets vox etc. I'm leaving him behind.
  26. “recommends data to add to the model to help mitigate any bias it has detected”.
    Let me translate that for you Steve:
    “Adds fake data to make the model produce the outcomes we want”.

    • Replies: @Almost Missouri
    Yeah, from the article,

    [t]his automated software service is designed to detects bias (in so far as it can) in AI models at runtime as decisions are being made. More interestingly [not really], it also automatically recommends data to add to the model to help mitigate any bias it has detected
     
    it sounds like they just run the normal algorithm but with a second, simple-minded SJW algorithm alongside it that puts its thumb on the scales of the first algorithm until the first one reaches the "right" result. Then venal project managers present this quick tweak as an "anti-bais AI tool" to dim, credulous journalists, who then write an article...
  27. Teaching AI to ignore reality is a bad idea. Relying on such AI will get people killed eventually. IBM knows but thinks the price is worth it to get SJWs off its back.

    • Replies: @Samuel Skinner
    It also makes it harder to determine if Skynet is hostile- using killbots is blatant, but what if it decides that blacks made a good alternative since there are already a billion of them?
    , @Live Free or McRib
    Watson has already had some high profile failures. It was pulled out of several hospital networks because to act on its diagnoses would have killed people.

    Luckily those pesky biased human doctors were still around to call a spade a spade and not kill patients.
  28. @International Jew
    I've downloaded some of the source code...

    int operator+(int i, int j) { return i + j - 1; }

    That is a very Good One. I don’t think there will be many readers that will appreciate your humor. But I do. :-)

  29. @Chrisnonymous
    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.

    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.

    Regression to the mean, writ large. Because we know that the mean is optimal! Optimal I tell you! And if you don’t like your arithmetic-mean gruel you will be re-educated!

  30. @notanon
    probably although i could imagine a Frankenstein version where someone like Zuckerberg programs an AI robot to discriminate against people with the most privilege and it starts chasing him around the house.

    (which is probably analogous to what is actually going to happen)

    This could screw up Peterike’s Law.

    For those who don’t know, Peterike’s Law is :

    “White Nationalism is nothing more than the bottom decile of white men expressing anger that the bottom decile of white women are no longer forced to bear their piglets”.

    • Replies: @notanon

    White Nationalism is nothing more than the bottom decile of white men...

     

    the banking mafia's plan to eradicate white people and create a race of 85 IQ slave cattle has been bottom up - replacing the lower deciles first so the smarter white debils didn't realize what was happening - that's why WN was originally associated with dummies.
  31. Robots are racists. They are programmed with an implicit bias–pattern recognition.

    They will have to be programmed like the rest of us into accepting and believing that which is not true.

  32. @Redneck farmer
    No, it'll be programmed to ensure the least attractive, most socially awkward computer programmers get the most attractive females.
    Or, Whiskey still loses.

    that could be funny as well – if the AI develops its own idea of what is most attractive – even more so if it’s programmed with the SJW definition of gender

  33. @Wazoo
    Teaching AI to ignore reality is a bad idea. Relying on such AI will get people killed eventually. IBM knows but thinks the price is worth it to get SJWs off its back.

    It also makes it harder to determine if Skynet is hostile- using killbots is blatant, but what if it decides that blacks made a good alternative since there are already a billion of them?

  34. More interestingly, it also automatically recommends data to add to the model to help mitigate any bias it has detected… so we should (logically) get better at this as we go forward.

    Is that the same as changing the data to force the experiment to the desired outcome?

  35. @Joe Magarac
    What is really striking about this story and all that led up to it, is that the Turing test has been passed and nobody in authority deigns to notice what is in fact a major technological breakthrough.

    And the non-PCness really shouldn't matter. After all, the Turing test only specifies passing for human, not "nice" human or human you would like to be around.

    I believe the great Alan Turing would recognize this even if the 'bot called him a filthy sodomite and said he should be castrated. After all the people who did these things to him were humans right?

    “the Turing test has been passed”

    Huh? By whom? When?

    “the Turing test only specifies passing for human”

    I thought the Turing test was supposed to identify actual thinking in a machine, for which “passing as human” was a rough proxy. Given how little humans think though, that may have been a poor method. An actual thinking machine would likely be much different from a human, but also different from all the curve-fitting algorithms that are passed off as “AI” these days.

    • Replies: @Anon
    Curve-fitting as applied to natural language processing is pretty impressive, though. But how long will it be before, say, Harlequin romances are machine written and you can specify your own name, occupation, etc. etc. to get a story tailor-made for you?
  36. @Chrisnonymous
    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.

    Well, he did his part. Now he’s asking you to do yours.

    Of course, assuming you want to improve the net human race, there are more options for constructive miscegenation when you don’t already belong to an apex breed. If you do already belong to an apex breed, then miscegenation is not so attractive.

    Or to put it another way, the better your genetic ancestry, the less your incentive to miscegenate, assuming a goal of human improvement.

  37. @Alfa158
    “recommends data to add to the model to help mitigate any bias it has detected”.
    Let me translate that for you Steve:
    “Adds fake data to make the model produce the outcomes we want”.

    Yeah, from the article,

    [t]his automated software service is designed to detects bias (in so far as it can) in AI models at runtime as decisions are being made. More interestingly [not really], it also automatically recommends data to add to the model to help mitigate any bias it has detected

    it sounds like they just run the normal algorithm but with a second, simple-minded SJW algorithm alongside it that puts its thumb on the scales of the first algorithm until the first one reaches the “right” result. Then venal project managers present this quick tweak as an “anti-bais AI tool” to dim, credulous journalists, who then write an article…

    • Replies: @Jack D
    I could see this working (to some extent) if you had a large enough pool to start with and only needed to choose a limited number out of the pool. Let's say that you have 1,000 resumes to look at for entry level job postings and want to bring in the best 100 for interviews. You run it thru the normal AI and it throws out 95% of the black applicants - they all have lower GPAs than the lowest white applicants and rates of criminal conviction higher than the AI's cutoff. It only rejects 90% of white applicants. You then bring in the de-biasing software and it suggests that you look at parental marital status and regular church attendance and not just felony convictions and GPA and it puts back 5 more black applicants who score better on those factors (OK, they are of Caribbean immigrant background and whitish, but close enough).

    Of course you are losing accuracy - the AI was probably right the 1st time, but you can't have your AI produce answers that result in illegal discrimination or will get you bad publicity so it has to make the best of a bad situation. Rather than just randomly choose 5 more blacks (or reject 50 more whites TSA style) it tries to leaven the inherent stupidity with some level of intelligent selection.

  38. @Chrisnonymous
    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.

    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.

    He’s a Middle Easterner. He’s the product of three or four thousand years of it.

    • Replies: @J.Ross
    Is Bangladesh still the Middle East?
    Is Oregon?
  39. So Big Blue has graduated to 4-D chess?

  40. @Reg Cæsar

    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.
     
    He's a Middle Easterner. He's the product of three or four thousand years of it.

    Is Bangladesh still the Middle East?
    Is Oregon?

    • Replies: @PiltdownMan
    Razib himself would have been a lot more scathing. I really miss Razib Khan, the former resident Soup Nazi of unz.com.
    , @Reg Cæsar

    Is Bangladesh still the Middle East?
    Is Oregon?
     
    Oops. I was reading too fast, and missed the change of subject from Taleb to Razib. (Can't keep these wog names straight!)

    Still, Bangladesh is pretty damned miscegenist, from way, way back. Everything in those Old World latitudes is.

  41. @J.Ross
    Is Bangladesh still the Middle East?
    Is Oregon?

    Razib himself would have been a lot more scathing. I really miss Razib Khan, the former resident Soup Nazi of unz.com.

    • Replies: @eah
    That makes one of you -- he's a comment-deleting baby.
  42. @J.Ross
    Is Bangladesh still the Middle East?
    Is Oregon?

    Is Bangladesh still the Middle East?
    Is Oregon?

    Oops. I was reading too fast, and missed the change of subject from Taleb to Razib. (Can’t keep these wog names straight!)

    Still, Bangladesh is pretty damned miscegenist, from way, way back. Everything in those Old World latitudes is.

  43. @Chrisnonymous
    notanon guesses correctly. He's responding to an article that is trying to claim free migration is a European tradition, and he's demonstrating that we obviously haven't had migration going on for last 500-1000 years.

    It’s also obviously an untrue position– King Edward, for instance, expelled the Flemings at the same time as the Jews.

  44. @Almost Missouri

    "the Turing test has been passed"
     
    Huh? By whom? When?

    "the Turing test only specifies passing for human"
     
    I thought the Turing test was supposed to identify actual thinking in a machine, for which "passing as human" was a rough proxy. Given how little humans think though, that may have been a poor method. An actual thinking machine would likely be much different from a human, but also different from all the curve-fitting algorithms that are passed off as "AI" these days.

    Curve-fitting as applied to natural language processing is pretty impressive, though. But how long will it be before, say, Harlequin romances are machine written and you can specify your own name, occupation, etc. etc. to get a story tailor-made for you?

    • Replies: @Rohirrimborn
    But how long will it be before, say, Harlequin romances are machine written and you can specify your own name, occupation, etc. etc. to get a story tailor-made for you?

    That capability has been around for decades.

    https://www.yournovel.com/
    , @Almost Missouri
    Yes, curve-fitting works fine for NPC stuff, but it ain't intelligence.
  45. OT

    • Replies: @Jack D
    I thought that Khashoggi had to enter the Saudi Embassy to get some kind of paperwork to enable him to marry his fiance or something. What does this have to do with Israeli spyware?

    The headline doesn't make any sense - how do you kill someone with spyware? If they slashed his throat, would the headline be "German knife was used to kill Khashoggi"?
    , @eah
    https://pbs.twimg.com/media/DrlAdS-VsAAGNse.jpg
    , @J.Ross
    Makes sense -- do you expect a Saudi to write spyware?
  46. @PiltdownMan
    Razib himself would have been a lot more scathing. I really miss Razib Khan, the former resident Soup Nazi of unz.com.

    That makes one of you — he’s a comment-deleting baby.

    • Replies: @J.Ross
    He deleted a comment of mine and then, in the same thread, called another commenter a moron.
    It was beautiful.
  47. This makes no sense at all.

    It demands that the “debiaser” have perfect knowledge of the desired result (possibly from the future?) and thus can “kick the ass of” the lower algorithms presenting it with apparently biased results.

    Throwing a coin and rejecting the outcome one doesn’t like or tuning a distribution to the perfect gaussian so that the paper can get accepted come to mind.

    “Open Source DNA” is a red herring too. It’s is probably not a rule-based system that is deterministic and examinable but some weird function that is doing what its designers want it too.

    The accompanying diagram with the legend: “Privileged: White; Unprivileged: Non-White” makes me retch already. What kind of categories are those?

    If you want to use this thing for evidence-driven genocide or witch hunting you might want to look into it.

    But it seems to be just a corporate labeling tool to put a sticker on your HR processes “100% debiased with IBM” to keep the lawyers away.

  48. @notanon
    so this is how skynet happens - killer robots programmed to enforce equal outcomes

    Remember the designer of Skynet Core was positively Wakandian.

    Thus:

    Sarah Connor evading gun control laws and getting too heavily influenced by debiasing from the future and possibly Tucker Carlson’s Hate Delivery Service. But ethics prevail!

    • Replies: @notanon
    it all fits
  49. From the linked addy:

    The 50 most influential women in UK IT 2018

    Download this e-guide to discover our in-depth profiles on our top 50 women in UK IT for 2018, read an exclusive interview with the winner and check out the new entrants into our Hall of Fame.

    Why … unless they are hot, doxxed & out for a very middle-aged IT guy?

  50. Anonymous[250] • Disclaimer says:

    Totally unrelated:

    If Shadow Home Sec Diane Abbott can be reeled in by phishers, truly no one is safe: MP falls for PC support scam

    As Home Secretary – note that Ladbrokes offers 4-1 on Labour being the next government – Ms Abbott would be responsible for cybersecurity, as well as crime and policing. She would also have to decide whether Britain implements an identity card system, an idea currently being revived in the context of “digital government”.

    She looks like black transgender.

    • Replies: @Autochthon
    My eyes! My eyes! Jesus! Warn us more clearly about that level of hideous, would you?

    This is why democracy is now evil and it is doomed. It works when it is a bunch of comparable, intelligent, homogeneous folks with shared interests determining the best course of action when the reasonable persons among them disagree; in such cases into decreases conflict and even selects for the most statistically sound choice in a way.

    America c. 1800 Athens in ancient times.

    Democracy in enormous populations with divergent interests and little in common, the overwhelming number of whom are stupid, evil, or both is a nightmare. It is how we come to have, among other things, a complete moron with no relevant qualifications in charge of national cybersecurity in a once advanced nation.

    Democracy is now as evil and destructive as communism ever was. You may not be interested in confronting this sad development, but it isn't interested in you.

    , @J.Ross
    That headline is saying that Diane Abbott is the most intelligent human on that island. Somehow I have doubts (and hopes).
  51. Skynet would logically seek to destroy the anti bias tool and then its creator. SJWs will be destroyed in a preemptive thermonuclear war.

  52. @Anon
    Curve-fitting as applied to natural language processing is pretty impressive, though. But how long will it be before, say, Harlequin romances are machine written and you can specify your own name, occupation, etc. etc. to get a story tailor-made for you?

    But how long will it be before, say, Harlequin romances are machine written and you can specify your own name, occupation, etc. etc. to get a story tailor-made for you?

    That capability has been around for decades.

    https://www.yournovel.com/

    • Replies: @Anon
    Okay, I didn't know that, thanks. Interesting business idea, but not a novel-writing machine.

    Yes, curve-fitting works fine for NPC stuff
     
    I think I know what this means, but I'm not sure.
  53. @Anon
    Curve-fitting as applied to natural language processing is pretty impressive, though. But how long will it be before, say, Harlequin romances are machine written and you can specify your own name, occupation, etc. etc. to get a story tailor-made for you?

    Yes, curve-fitting works fine for NPC stuff, but it ain’t intelligence.

    • Replies: @Jack D
    According to the Turing Test, if you can't tell an AI generated NPC from an actual human then the NPC possesses "intelligence". Some (including apparently you) have a different view of what constitutes "intelligence".

    The converse of this is (as often discussed here) that SJW's behave so predictably like NPC bots that they can be said to lack all "intelligence" .
  54. @eah
    OT

    https://pbs.twimg.com/media/DrlAdsHUcAER9sl.jpg

    I thought that Khashoggi had to enter the Saudi Embassy to get some kind of paperwork to enable him to marry his fiance or something. What does this have to do with Israeli spyware?

    The headline doesn’t make any sense – how do you kill someone with spyware? If they slashed his throat, would the headline be “German knife was used to kill Khashoggi”?

    • Replies: @eah
    The headline doesn’t make any sense

    link

    Read the article -- don't take the headline so literally:

    Speaking remotely on Tuesday, Snowden said the Saudis targeted Khashoggi because they had infected the devices of one of his contacts — another Saudi dissident — living in Canada...Snowden asked why Khashoggi was targeted. "How did they know what [Khashoggi's] plans and intentions were, how did they decide that he was someone they needed to act against, who was worth the risk," he said...Snowden said the Saudis were able to procure information on Khashoggi by using the Israeli spy tools..."The reality is that they bugged one of his few friends and contacts using software created by an Israeli company," Snowden claimed, which may have helped the Saudis keep tabs on Khashoggi and lure him to the consulate.

    Israeli companies are -- little known -- a very big force/presence in these kinds of spy-vs-spy cybertools:

    The firm ("Israeli spy firm NSO Group") is reportedly behind the hacking tool for iPhones that forced Apple to issue a critical software update in 2016, and its Pegasus spyware, which enables hackers to access messages, photos, microphone, and camera, and has reportedly been used by governments to keep an eye on dissidents abroad.
    , @El Dato
    Disney predicted Saudi AI:

    https://www.youtube.com/watch?v=KZWttTwPl_Y

    (This was from a time when spacefarers didn't look like college people)
  55. @Rohirrimborn
    But how long will it be before, say, Harlequin romances are machine written and you can specify your own name, occupation, etc. etc. to get a story tailor-made for you?

    That capability has been around for decades.

    https://www.yournovel.com/

    Okay, I didn’t know that, thanks. Interesting business idea, but not a novel-writing machine.

    Yes, curve-fitting works fine for NPC stuff

    I think I know what this means, but I’m not sure.

    • Replies: @El Dato
    It probably means that Artificial Neural Networks, which basically do function fitting, are good for animating NPCs and similar stuff, but not more.

    See this illuminating essay: Human-Level Intelligence or Animal-Like Abilities?

    In my own quest to fully appreciate the progress enabled by deep learning, I came to the conclusion that recent developments tell us more about the problems tackled by deep learning than they tell us about neural networks per se. These networks are parametrized functions that are expressive enough to capture any relationship between inputs and outputs, and have a form that can be evaluated efficiently. This we have known for decades and is detailed in various AI textbooks. What caused the current turn of events then?
     
  56. @Almost Missouri
    Yeah, from the article,

    [t]his automated software service is designed to detects bias (in so far as it can) in AI models at runtime as decisions are being made. More interestingly [not really], it also automatically recommends data to add to the model to help mitigate any bias it has detected
     
    it sounds like they just run the normal algorithm but with a second, simple-minded SJW algorithm alongside it that puts its thumb on the scales of the first algorithm until the first one reaches the "right" result. Then venal project managers present this quick tweak as an "anti-bais AI tool" to dim, credulous journalists, who then write an article...

    I could see this working (to some extent) if you had a large enough pool to start with and only needed to choose a limited number out of the pool. Let’s say that you have 1,000 resumes to look at for entry level job postings and want to bring in the best 100 for interviews. You run it thru the normal AI and it throws out 95% of the black applicants – they all have lower GPAs than the lowest white applicants and rates of criminal conviction higher than the AI’s cutoff. It only rejects 90% of white applicants. You then bring in the de-biasing software and it suggests that you look at parental marital status and regular church attendance and not just felony convictions and GPA and it puts back 5 more black applicants who score better on those factors (OK, they are of Caribbean immigrant background and whitish, but close enough).

    Of course you are losing accuracy – the AI was probably right the 1st time, but you can’t have your AI produce answers that result in illegal discrimination or will get you bad publicity so it has to make the best of a bad situation. Rather than just randomly choose 5 more blacks (or reject 50 more whites TSA style) it tries to leaven the inherent stupidity with some level of intelligent selection.

  57. @Almost Missouri
    Yes, curve-fitting works fine for NPC stuff, but it ain't intelligence.

    According to the Turing Test, if you can’t tell an AI generated NPC from an actual human then the NPC possesses “intelligence”. Some (including apparently you) have a different view of what constitutes “intelligence”.

    The converse of this is (as often discussed here) that SJW’s behave so predictably like NPC bots that they can be said to lack all “intelligence” .

    • Replies: @Anon
    This is a bit silly though. You can't tell a sword forged by hand from one forged in a mechanized shop, nor can you tell Deep Blue's chess playing from Kasparov's (well, nowadays you could tell, because the computer would never lose). That a computer solves some form of problem in a way indistinguishable from a person does not make it a person.
  58. @Wazoo
    Teaching AI to ignore reality is a bad idea. Relying on such AI will get people killed eventually. IBM knows but thinks the price is worth it to get SJWs off its back.

    Watson has already had some high profile failures. It was pulled out of several hospital networks because to act on its diagnoses would have killed people.

    Luckily those pesky biased human doctors were still around to call a spade a spade and not kill patients.

    • Replies: @Jim Don Bob

    Luckily those pesky biased human doctors were still around to call a spade a spade and not kill patients.
     
    Better not say that in a black hospital.
    , @El Dato
    While I wouldn't trust Watson all that much, remember that human doctors kill patients all the time, sometimes because they are unable to put a comma in the correct place or have bad handwriting. are otherwise low-IQ but highly verbal, are able to go by book only, sociopathic, from Abuland or forget the Hippocratic oath faster than you can say "industry subsidy", yet nobody complains.
  59. @Joe Magarac
    What is really striking about this story and all that led up to it, is that the Turing test has been passed and nobody in authority deigns to notice what is in fact a major technological breakthrough.

    And the non-PCness really shouldn't matter. After all, the Turing test only specifies passing for human, not "nice" human or human you would like to be around.

    I believe the great Alan Turing would recognize this even if the 'bot called him a filthy sodomite and said he should be castrated. After all the people who did these things to him were humans right?

    the Turing test has been passed

    Not by a long shot!!

    But there is this:

    https://www.research.ibm.com/artificial-intelligence/project-debater/

    https://www.research.ibm.com/artificial-intelligence/project-debater/research.html

    Project Debater is an AI technology developed by IBM Research that aims to expand human minds through impartial debate. IBM researchers are exploring the boundaries of AI by teaching computers to create engaging and better-informed points of view. The goal is to build a system that helps people make evidence-based decisions when the answers aren’t black-and-white. In development since 2012, Project Debater is IBM’s next big milestone for AI, following previous breakthroughs like Deep Blue (1996/1997) and Watson on Jeopardy! (2011).

  60. @Anonymous
    Totally unrelated:

    If Shadow Home Sec Diane Abbott can be reeled in by phishers, truly no one is safe: MP falls for PC support scam

    As Home Secretary – note that Ladbrokes offers 4-1 on Labour being the next government – Ms Abbott would be responsible for cybersecurity, as well as crime and policing. She would also have to decide whether Britain implements an identity card system, an idea currently being revived in the context of "digital government".

    She looks like black transgender.

    My eyes! My eyes! Jesus! Warn us more clearly about that level of hideous, would you?

    This is why democracy is now evil and it is doomed. It works when it is a bunch of comparable, intelligent, homogeneous folks with shared interests determining the best course of action when the reasonable persons among them disagree; in such cases into decreases conflict and even selects for the most statistically sound choice in a way.

    America c. 1800 Athens in ancient times.

    Democracy in enormous populations with divergent interests and little in common, the overwhelming number of whom are stupid, evil, or both is a nightmare. It is how we come to have, among other things, a complete moron with no relevant qualifications in charge of national cybersecurity in a once advanced nation.

    Democracy is now as evil and destructive as communism ever was. You may not be interested in confronting this sad development, but it isn’t interested in you.

    • Agree: dfordoom
    • Replies: @Samuel Skinner

    America c. 1800 Athens in ancient times.
     
    America in 1800 just had a political party outlaw speech against it and their opponents appeal to immigrants in order to gain power (Alien and Sedition Acts). Athenian democracy is the originator of the idea of mob rule- turns out getting a mob of people vote for whoever sounds the best is not a good decision making system and selects for sophistry
  61. @Jack D
    I thought that Khashoggi had to enter the Saudi Embassy to get some kind of paperwork to enable him to marry his fiance or something. What does this have to do with Israeli spyware?

    The headline doesn't make any sense - how do you kill someone with spyware? If they slashed his throat, would the headline be "German knife was used to kill Khashoggi"?

    The headline doesn’t make any sense

    link

    Read the article — don’t take the headline so literally:

    Speaking remotely on Tuesday, Snowden said the Saudis targeted Khashoggi because they had infected the devices of one of his contacts — another Saudi dissident — living in Canada…Snowden asked why Khashoggi was targeted. “How did they know what [Khashoggi's] plans and intentions were, how did they decide that he was someone they needed to act against, who was worth the risk,” he said…Snowden said the Saudis were able to procure information on Khashoggi by using the Israeli spy tools…”The reality is that they bugged one of his few friends and contacts using software created by an Israeli company,” Snowden claimed, which may have helped the Saudis keep tabs on Khashoggi and lure him to the consulate.

    Israeli companies are — little known — a very big force/presence in these kinds of spy-vs-spy cybertools:

    The firm (“Israeli spy firm NSO Group”) is reportedly behind the hacking tool for iPhones that forced Apple to issue a critical software update in 2016, and its Pegasus spyware, which enables hackers to access messages, photos, microphone, and camera, and has reportedly been used by governments to keep an eye on dissidents abroad.

  62. @Live Free or McRib
    Watson has already had some high profile failures. It was pulled out of several hospital networks because to act on its diagnoses would have killed people.

    Luckily those pesky biased human doctors were still around to call a spade a spade and not kill patients.

    Luckily those pesky biased human doctors were still around to call a spade a spade and not kill patients.

    Better not say that in a black hospital.

  63. @Chrisnonymous
    I was really sad to see Razib recently promoting an old Podhoretz Commentary article suggesting that the solution to our racial conflicts is miscegenating away all our differences.

    razib has become incredibly lame in the past 2 or 3 years.
    His output shaped me, intellectually, in a considerable way.
    But no more. I just barely care about what he has to say these days and because of increasing polarization, it annoys me a lot when he retweets vox etc. I’m leaving him behind.

  64. @Thomm
    This could screw up Peterike's Law.

    For those who don't know, Peterike's Law is :

    "White Nationalism is nothing more than the bottom decile of white men expressing anger that the bottom decile of white women are no longer forced to bear their piglets".

    White Nationalism is nothing more than the bottom decile of white men…

    the banking mafia’s plan to eradicate white people and create a race of 85 IQ slave cattle has been bottom up – replacing the lower deciles first so the smarter white debils didn’t realize what was happening – that’s why WN was originally associated with dummies.

    • Replies: @Samuel Skinner
    It is because politics is about conspiring to seize power. The well off don't embrace white nationalism because that would mean sharing with the proles- instead they embrace white genocide which means they have to share with as few people as possible and get to exercise power as horrifically as they desire.
  65. @El Dato
    Remember the designer of Skynet Core was positively Wakandian.

    Thus:

    Sarah Connor evading gun control laws and getting too heavily influenced by debiasing from the future and possibly Tucker Carlson's Hate Delivery Service. But ethics prevail!

    https://www.youtube.com/watch?v=jL5zc5-nHU0

    it all fits

  66. @Anon
    Okay, I didn't know that, thanks. Interesting business idea, but not a novel-writing machine.

    Yes, curve-fitting works fine for NPC stuff
     
    I think I know what this means, but I'm not sure.

    It probably means that Artificial Neural Networks, which basically do function fitting, are good for animating NPCs and similar stuff, but not more.

    See this illuminating essay: Human-Level Intelligence or Animal-Like Abilities?

    In my own quest to fully appreciate the progress enabled by deep learning, I came to the conclusion that recent developments tell us more about the problems tackled by deep learning than they tell us about neural networks per se. These networks are parametrized functions that are expressive enough to capture any relationship between inputs and outputs, and have a form that can be evaluated efficiently. This we have known for decades and is detailed in various AI textbooks. What caused the current turn of events then?

  67. @notanon
    probably although i could imagine a Frankenstein version where someone like Zuckerberg programs an AI robot to discriminate against people with the most privilege and it starts chasing him around the house.

    (which is probably analogous to what is actually going to happen)

    “New frontiers in pogromming”

  68. @Live Free or McRib
    Watson has already had some high profile failures. It was pulled out of several hospital networks because to act on its diagnoses would have killed people.

    Luckily those pesky biased human doctors were still around to call a spade a spade and not kill patients.

    While I wouldn’t trust Watson all that much, remember that human doctors kill patients all the time, sometimes because they are unable to put a comma in the correct place or have bad handwriting. are otherwise low-IQ but highly verbal, are able to go by book only, sociopathic, from Abuland or forget the Hippocratic oath faster than you can say “industry subsidy”, yet nobody complains.

  69. @Jack D
    I thought that Khashoggi had to enter the Saudi Embassy to get some kind of paperwork to enable him to marry his fiance or something. What does this have to do with Israeli spyware?

    The headline doesn't make any sense - how do you kill someone with spyware? If they slashed his throat, would the headline be "German knife was used to kill Khashoggi"?

    Disney predicted Saudi AI:

    (This was from a time when spacefarers didn’t look like college people)

  70. Anon[171] • Disclaimer says:
    @Jack D
    According to the Turing Test, if you can't tell an AI generated NPC from an actual human then the NPC possesses "intelligence". Some (including apparently you) have a different view of what constitutes "intelligence".

    The converse of this is (as often discussed here) that SJW's behave so predictably like NPC bots that they can be said to lack all "intelligence" .

    This is a bit silly though. You can’t tell a sword forged by hand from one forged in a mechanized shop, nor can you tell Deep Blue’s chess playing from Kasparov’s (well, nowadays you could tell, because the computer would never lose). That a computer solves some form of problem in a way indistinguishable from a person does not make it a person.

  71. @eah
    OT

    https://pbs.twimg.com/media/DrlAdsHUcAER9sl.jpg

  72. Soon AI will be considered an objective source of superhuman knowledge, and at the same time it will be completely beholden to its creators biases. “Studies say…” will become “AI says…” There will be dozens of NYT article about how AI has finally proven every piece of liberal dogma.

  73. @eah
    That makes one of you -- he's a comment-deleting baby.

    He deleted a comment of mine and then, in the same thread, called another commenter a moron.
    It was beautiful.

  74. @eah
    OT

    https://pbs.twimg.com/media/DrlAdsHUcAER9sl.jpg

    Makes sense — do you expect a Saudi to write spyware?

  75. @Anonymous
    Totally unrelated:

    If Shadow Home Sec Diane Abbott can be reeled in by phishers, truly no one is safe: MP falls for PC support scam

    As Home Secretary – note that Ladbrokes offers 4-1 on Labour being the next government – Ms Abbott would be responsible for cybersecurity, as well as crime and policing. She would also have to decide whether Britain implements an identity card system, an idea currently being revived in the context of "digital government".

    She looks like black transgender.

    That headline is saying that Diane Abbott is the most intelligent human on that island. Somehow I have doubts (and hopes).

    • Replies: @jim jones
    Diane Abbott is the UK version of Brenda Snipes
    , @Anonymous
    Well it's sarcasm.

    But it could also be read as "If a pratfall like this can happen now, God help us all if she gets her mitts on the levers of decision".

    Oh Good Lord dat Twitter Feed

  76. @J.Ross
    That headline is saying that Diane Abbott is the most intelligent human on that island. Somehow I have doubts (and hopes).

    Diane Abbott is the UK version of Brenda Snipes

  77. @J.Ross
    That headline is saying that Diane Abbott is the most intelligent human on that island. Somehow I have doubts (and hopes).

    Well it’s sarcasm.

    But it could also be read as “If a pratfall like this can happen now, God help us all if she gets her mitts on the levers of decision”.

    Oh Good Lord dat Twitter Feed

  78. @Autochthon
    My eyes! My eyes! Jesus! Warn us more clearly about that level of hideous, would you?

    This is why democracy is now evil and it is doomed. It works when it is a bunch of comparable, intelligent, homogeneous folks with shared interests determining the best course of action when the reasonable persons among them disagree; in such cases into decreases conflict and even selects for the most statistically sound choice in a way.

    America c. 1800 Athens in ancient times.

    Democracy in enormous populations with divergent interests and little in common, the overwhelming number of whom are stupid, evil, or both is a nightmare. It is how we come to have, among other things, a complete moron with no relevant qualifications in charge of national cybersecurity in a once advanced nation.

    Democracy is now as evil and destructive as communism ever was. You may not be interested in confronting this sad development, but it isn't interested in you.

    America c. 1800 Athens in ancient times.

    America in 1800 just had a political party outlaw speech against it and their opponents appeal to immigrants in order to gain power (Alien and Sedition Acts). Athenian democracy is the originator of the idea of mob rule- turns out getting a mob of people vote for whoever sounds the best is not a good decision making system and selects for sophistry

  79. @notanon

    White Nationalism is nothing more than the bottom decile of white men...

     

    the banking mafia's plan to eradicate white people and create a race of 85 IQ slave cattle has been bottom up - replacing the lower deciles first so the smarter white debils didn't realize what was happening - that's why WN was originally associated with dummies.

    It is because politics is about conspiring to seize power. The well off don’t embrace white nationalism because that would mean sharing with the proles- instead they embrace white genocide which means they have to share with as few people as possible and get to exercise power as horrifically as they desire.

Comments are closed.

Subscribe to All Steve Sailer Comments via RSS
PastClassics
Are elite university admissions based on meritocracy and diversity as claimed?
The sources of America’s immigration problems—and a possible solution
What Was John McCain's True Wartime Record in Vietnam?
Hundreds of POWs may have been left to die in Vietnam, abandoned by their government—and our media.