Skip to content
  • Home Home Home, current page.
  • Moments Moments Moments, current page.

Saved searches

  • Remove
  • In this conversation
    Verified accountProtected Tweets @
Suggested users
  • Verified accountProtected Tweets @
  • Verified accountProtected Tweets @
  • Language: English
    • Bahasa Indonesia
    • Bahasa Melayu
    • Català
    • Čeština
    • Dansk
    • Deutsch
    • English UK
    • Español
    • Filipino
    • Français
    • Hrvatski
    • Italiano
    • Magyar
    • Nederlands
    • Norsk
    • Polski
    • Português
    • Română
    • Slovenčina
    • Suomi
    • Svenska
    • Tiếng Việt
    • Türkçe
    • Ελληνικά
    • Български език
    • Русский
    • Српски
    • Українська мова
    • עִבְרִית
    • العربية
    • فارسی
    • मराठी
    • हिन्दी
    • বাংলা
    • ગુજરાતી
    • தமிழ்
    • ಕನ್ನಡ
    • ภาษาไทย
    • 한국어
    • 日本語
    • 简体中文
    • 繁體中文
  • Have an account? Log in
    Have an account?
    · Forgot password?

    New to Twitter?
    Sign up
zeynep's profile
zeynep tufekci
zeynep tufekci
zeynep tufekci
Verified account
@zeynep

Tweets

zeynep tufekciVerified account

@zeynep

Complex systems, wicked problems. Society, technology, science and more. @UNC professor. @NYTimes columnist. My newsletter is @insight: http://www.theinsight.org 

floating in a most peculiar way
theinsight.org
Joined August 2009

Tweets

  • © 2022 Twitter
  • About
  • Help Center
  • Terms
  • Privacy policy
  • Cookies
  • Ads info
Dismiss
Previous
Next

Go to a person's profile

Saved searches

  • Remove
  • In this conversation
    Verified accountProtected Tweets @
Suggested users
  • Verified accountProtected Tweets @
  • Verified accountProtected Tweets @

Promote this Tweet

Block

  • Tweet with a location

    You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more

    Your lists

    Create a new list


    Under 100 characters, optional

    Privacy

    Copy link to Tweet

    Embed this Tweet

    Embed this Video

    Add this Tweet to your website by copying the code below. Learn more

    Add this video to your website by copying the code below. Learn more

    Hmm, there was a problem reaching the server.

    By embedding Twitter content in your website or app, you are agreeing to the Twitter Developer Agreement and Developer Policy.

    Preview

    Why you're seeing this ad

    Log in to Twitter

    · Forgot password?
    Don't have an account? Sign up »

    Sign up for Twitter

    Not on Twitter? Sign up, tune into the things you care about, and get updates as they happen.

    Sign up
    Have an account? Log in »

    Two-way (sending and receiving) short codes:

    Country Code For customers of
    United States 40404 (any)
    Canada 21212 (any)
    United Kingdom 86444 Vodafone, Orange, 3, O2
    Brazil 40404 Nextel, TIM
    Haiti 40404 Digicel, Voila
    Ireland 51210 Vodafone, O2
    India 53000 Bharti Airtel, Videocon, Reliance
    Indonesia 89887 AXIS, 3, Telkomsel, Indosat, XL Axiata
    Italy 4880804 Wind
    3424486444 Vodafone
    » See SMS short codes for other countries

    Confirmation

     

    Welcome home!

    This timeline is where you’ll spend most of your time, getting instant updates about what matters to you.

    Tweets not working for you?

    Hover over the profile pic and click the Following button to unfollow any account.

    Say a lot with a little

    When you see a Tweet you love, tap the heart — it lets the person who wrote it know you shared the love.

    Spread the word

    The fastest way to share someone else’s Tweet with your followers is with a Retweet. Tap the icon to send it instantly.

    Join the conversation

    Add your thoughts about any Tweet with a Reply. Find a topic you’re passionate about, and jump right in.

    Learn the latest

    Get instant insight into what people are talking about now.

    Get more of what you love

    Follow more accounts to get instant updates about topics you care about.

    Find what's happening

    See the latest conversations about any topic instantly.

    Never miss a Moment

    Catch up instantly on the best stories happening as they unfold.

    1. Benedict Evans‏Verified account @benedictevans 18 Nov 2018
      Replying to @zeynep @stevesi @qhardy

      And so are all flawed institutions. None of this supports your basic assertion that the risk of misuse or misunderstanding of ML is different in principle from the ways all other techs & processes are subject to misuse or misunderstanding. ‘It’s not auditable’ is not good enough

      3 replies 0 retweets 3 likes
    2. zeynep tufekci‏Verified account @zeynep 18 Nov 2018
      Replying to @benedictevans @stevesi @qhardy

      ML is going to allow us to detect things at scale and cheaply that we could not before. That, in the hands of the powerful, can be a terrible tool. I can write the awesome scenarios but.. until recently, you just couldn't detect say, gay or rebel or uyghur, *at scale* and cheap.+

      3 replies 4 retweets 27 likes
    3. zeynep tufekci‏Verified account @zeynep 18 Nov 2018
      Replying to @zeynep @benedictevans and

      Plus, ML will allow us to classify and optimize at scale, and be better at it than humans potentially, but opaquely... Humans hire from alumni network, have gender/race biases in hiring and are credentialist. What is ML going to weed out? Don't even know where to begin to look.+

      2 replies 1 retweet 19 likes
    4. Benedict Evans‏Verified account @benedictevans 18 Nov 2018
      Replying to @zeynep @stevesi @qhardy

      You keep making this assertion. I keep pointing out why it’s flawed. This would be a more productive conversation if you could respond to that.

      2 replies 0 retweets 2 likes
    5. zeynep tufekci‏Verified account @zeynep 18 Nov 2018
      Replying to @benedictevans @stevesi @qhardy

      When we take formal rules and put them in a database (say the CA immigration system: it adds points. Doesn't matter by hand or computation) or even when we automate a fairly well-understood system (flying) we have ways of debugging/troubleshooting that we don't have for ML. +

      1 reply 0 retweets 5 likes
    6. zeynep tufekci‏Verified account @zeynep 18 Nov 2018
      Replying to @zeynep @benedictevans and

      So the Q is: do you think we will have explainable/interpretable AI? I'm on the this doesn't seem anymore likely than using brain scans to understand humans. ML is classification that works not via rules we wrote (not Symbolic/Minsky style code). That really is different! +

      2 replies 0 retweets 11 likes
    7. This Tweet is unavailable.
    8. Quentin Hardy‏Verified account @qhardy 18 Nov 2018
      Replying to @stevesi @zeynep @benedictevans

      One can also write a supra program that looks at results and tags, and reports anomalies, or things society deems are biases. Frequently, the problem with ML may be the mirror it holds up to us.

      1 reply 0 retweets 4 likes
    9. zeynep tufekci‏Verified account @zeynep 18 Nov 2018
      Replying to @qhardy @stevesi @benedictevans

      Only for obvious variables like race/gender which we know to look for. That's why there is so much reporting on that. Familiar ground. But ML will detect and discriminate things we could not previously detect, will not even think to check for. No variable list to run against.

      1 reply 3 retweets 22 likes
    10. zeynep tufekci‏Verified account @zeynep 18 Nov 2018
      Replying to @zeynep @qhardy and

      I think the current convo on AI/ML is what if it is wrong in the sense it reflects the biases in the training data which are often structural biases in human societies. I'm asking something different: what if it works, and detects something we previously could not—not at scale.

      1 reply 5 retweets 20 likes
      zeynep tufekci‏Verified account @zeynep 18 Nov 2018
      Replying to @zeynep @qhardy and

      And it's opacity means that we don't even know whatever latent thing it's using to classify. Instead of not hiring based on race, it's not hiring based on propensity to depression—and nobody knows that's what it's doing. No variable labeled "prop_depression". Just a giant matrix.

      5:21 PM - 18 Nov 2018
      • 28 Likes
      • Emily March Baris Kent Wayland J Emory Parker 🏳️‍🌈 Subscribe to STAT+ Reuben Binns Miss IG Geek (she/her) 🏳️‍🌈 Félix Cantournet
      5 replies 0 retweets 28 likes
        1. This Tweet is unavailable.
        2. grisaitis‏ @grisaitis 18 Nov 2018
          Replying to @stevesi @zeynep and

          Definitely with @zeynep on this one. FYI @_DennisLi @PhABCD @docrostov what do you think

          1 reply 0 retweets 0 likes
        3. Show replies
        1. Bill Stewart‏ @BillStewart415 18 Nov 2018
          Replying to @zeynep @qhardy and

          As a computer scientist, I differentiate between algorithms, which are supposed to be explainable, (even if "how you tweeked the parameters" isn't always), and ML which is heavily data-dependent, in ways that are usually opaque, because data's got lots of latent stuff in it.

          0 replies 0 retweets 3 likes
          Thanks. Twitter will use this to make your timeline better. Undo
          Undo
        1. New conversation
        2. Nicolas Saunier‏ @nicolas_saunier 18 Nov 2018
          Replying to @zeynep @qhardy and

          We do know the input to any ML model. The problem is that the decision may be correlated to such things we don't know we can detect. But it's not a black box, at least not in the sense that we cannot study it's response to known inputs and compare to expected outputs.

          1 reply 0 retweets 1 like
        3. zeynep tufekci‏Verified account @zeynep 18 Nov 2018
          Replying to @nicolas_saunier @qhardy and

          But you don't know which of the million outputs matter, how, and if they group and correspond to latent variables that you don't even know to look for because it's not an input by itself. The construct ML is using may not even yet exist in human understanding but be predictive.

          1 reply 0 retweets 1 like
        4. Show replies
        1. New conversation
        2. Krylon Sprayed Atoms‏ @KevinSheaAdams 18 Nov 2018
          Replying to @zeynep @qhardy and

          That there’s now also a naive or colloquial sense of “algorithm” doesn’t seem fruitful. I understand how ML is opaque but doubt it’s opacity will approach “human”. Hard to believe algorithmic biases will ever be so hidden or profound, though they may be immensely more destructive

          1 reply 0 retweets 0 likes
        3. Krylon Sprayed Atoms‏ @KevinSheaAdams 18 Nov 2018
          Replying to @KevinSheaAdams @zeynep and

          In other words, I’m sure ML biases will be able to reproduce complex racial prejudice and be discoverable as such. But I sincerely doubt they will opaquely start discovering something like “good art”.

          0 replies 0 retweets 0 likes
        4. End of conversation
        1. New conversation
        2. E. Mehmet Kıral‏ @EMehmetKiral 18 Nov 2018
          Replying to @zeynep @qhardy and

          Algorithms are scary even without the ML. Let me tell a story from Japan. We wanted to have internet banking in Japan. The bank had recently decided that this was only going to be possible through a one time password app. The app is available in Google play. But only in Japan.

          1 reply 0 retweets 1 like
        3. E. Mehmet Kıral‏ @EMehmetKiral 18 Nov 2018
          Replying to @EMehmetKiral @zeynep and

          Turns out, our Google play location is still registered in US (even though we have Japanese phones). You don't even see the change country button. We searched a while then deleted our us payment options, only then you could see the change country option. Can change once a year.

          1 reply 0 retweets 0 likes
        4. Show replies
        1. apotheosia‏ @joiedeverte 18 Nov 2018
          Replying to @zeynep @qhardy and

          I agree with you. ML as it currently exists doesn’t show its work enough to allow human troubleshooting and tinkering if and when society realizes that the effects of some ML decision-making are discriminatory or otherwise detrimental.

          0 replies 0 retweets 1 like
          Thanks. Twitter will use this to make your timeline better. Undo
          Undo

      Loading seems to be taking a while.

      Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

        Promoted Tweet

        false

        • © 2022 Twitter
        • About
        • Help Center
        • Terms
        • Privacy policy
        • Cookies
        • Ads info