Aylin Caliskan

@aylin_cim

professor & fellow interested in AI ethics, algorithmic bias, computer vision, machine learning, natural language processing

Seattle, WA
seas.gwu.edu/~aylin
Joined July 2013
19 Photos and videos Photos and videos

Tweets

You blocked @aylin_cim

Are you sure you want to view these Tweets? Viewing Tweets won't unblock @aylin_cim

  1. Pinned Tweet
    Jun 3

    I'm delighted to be moving to the University of Washington's Information School as an assistant professor to research ethics, bias, and equity in AI; after 3 wonderful years at George Washington University. Look forward to working with new colleagues & in Seattle.

    Aylin Caliksan
    Aylin Caliksan () joins us as an assistant professor focused on ethics and bias in , machine learning and the implications of machine intelligence for equity and fairness.
    Show this thread
    25 replies 3 retweets 286 likes
    Undo
  2. Retweeted
    11 hours ago

    Very excited to announce this paper and open dataset!! 🎉🎉Implicit and explicit attitude and stereotype data from 34 countries (plus two bilingual datasets!). Please share widely!!

    New alert! 🚨 by and me, w/ & Yoav Rabinovich: We present a unique and comprehensive dataset of 2.3 million tests across 7 attitude topics, using both indirect & direct measures. 1/n
    Show this thread
    2 retweets 16 likes
    Undo
  3. Retweeted
    9 hours ago

    The following are the planned dates for AIES'22: February 22: submission deadline April 18: notification Early August: conference ()

    6 retweets 19 likes
    Undo
  4. Retweeted
    13 hours ago

    JUST LAUNCHED: The AI Researchers Portal on . It’s a central connection to many Federally-supported resources for America’s AI research community. Click here for details on grant programs, testbeds, datasets, and much more:

    1 reply 62 retweets 137 likes
    Show this thread
    Show this thread
    Undo
  5. Retweeted
    Dec 15

    Good news! officially announced for August 2022 at Oxford!

    Apologies for the silence as we tried to figure out how to maximize the chance of a physical conference... We're planning to hold AIES'22 in Oxford in early August, hosted by . Deadline to follow shortly!
    1 retweet 4 likes
    Undo
  6. Retweeted
    Dec 14

    My research group has been following a cohort of ~50 gig workers for 2 years, and have interviewed many others -this figure from Pew's recent survey does not fit with our findings at all...If you gave me the distributions unlabeled I would put numbers exactly the other way around

    1 reply 2 retweets 6 likes
    Show this thread
    Show this thread
    Undo
  7. Retweeted
    Dec 13

    "Attention is all you need" implementation from scratch in PyTorch. A Twitter thread: 1/

    30 replies 615 retweets 3,135 likes
    Show this thread
    Show this thread
    Undo
  8. Retweeted
    Dec 12

    Make sure you submit your work to this track!

    Join us in this exciting new track at Track chairs: , , Amir Banifatemi, and me
    7 retweets 15 likes
    Undo
  9. Retweeted
    Dec 6

    Our new book, Critically Conscious Computing: Methods for Secondary Education, is live! 1/n Book: Blog post:

    5 replies 128 retweets 350 likes
    Show this thread
    Show this thread
    Undo
  10. Retweeted
    Dec 2

    Guys calm down, stop freaking out, it's just a bunch of actuators and motors with rubber stretched over them to simulate facial movement. It's all preprogrammed. It's just for research. Internally: nope nope nope nope nope

    0:41
    I don’t know man
    2,118 replies 10,053 retweets 81,503 likes
    Undo
  11. Retweeted
    Dec 2

    Almost six years ago, wrote an article using real data that showed how PredPol could exacerbate racial disparities in policing. 's release today shows that that is *exactly* what happened.

    Critics have long suspected that predictive policing software was racially biased. Today, we have the answer: & analyzed 5.9 million algorithmic crime predictions. We found they disproportionately target Black & Latino areas. /1
    Show this thread
    4 replies 63 retweets 162 likes
    Show this thread
    Show this thread
    Undo
  12. Retweeted
    Nov 29

    The combination of and capitalism works to maximize profits, sometimes with unintended consequences. The iSchool's Aylin Caliskan () recently went on 's In Machines We Trust podcast to discuss it.

    1 retweet 2 likes
    Undo
  13. Retweeted
    Nov 17

    Please join the Responsible AI Systems & Experiences (RAISE) group at the University of Washington for the talk “A Perspective on AI Governance” by Kush R. Varshney, a distinguished research staff member and manager with IBM Research, from 9-10am PST Friday, 12/3.

    1 retweet 6 likes
    Undo
  14. Retweeted
    Nov 12

    The new and improved version of "Data and its (dis)contents" is published at today! Co-authored with . Check it out here:

    Dataset design & development: Dataset audits reveal representational harms. Spurious cues exploited by models lead to unanticipated results. Dataset construction can legitimize faulty science. Datasets have been historically insufficiently documented and motivated. Dataset use: Meticulous, human inspection of large datasets turns up disturbing content. Automated dataset improvement is limited by validity of task definition and initial data collection. Dataset culture: Leaderboardism distorts the science of ML research. Data management practice lacks a culture of care for data subjects. Dataset appropriation and reuse practices break connections to context. The push for massive scale engenders poor labor conditions. Dataset collection practices raise legal issues and existing legal frameworks provide insufficient protection to data subjects.
    1 reply 42 retweets 120 likes
    Show this thread
    Show this thread
    Undo
  15. Retweeted
    Nov 10

    The bill passed 38-4! Congratulations to all and a big thank you to

    Int 1894 "in relation to automated employment decision tools" up for a vote at , with strong disclosure requirements: to notify job applicants that a tool will be used to screen them, allowing them to request an accommodation
    Show this thread
    3 retweets 16 likes
    Undo
  16. Nov 8

    Our new semantics evaluation method at Resources and Evaluation: ValNorm quantifies semantics to reveal consistent non-discriminatory, non-social group valence biases across languages and over centuries, whereas social group biases are evolving.

    2 replies 3 likes
    Undo
  17. Nov 8

    Our paper at 's Ethics and NLP session: underrepresentation of minority groups in language models results in bias and overfitting as analyzed w.r.t. frequency in the representation space of contextualized word embeddings.

    1 reply 5 likes
    Undo
  18. Retweeted
    Oct 31

    Excited that the final version of "Physiognomic Artificial Intelligence" will appear in the Fordham Intellectual Property, Media & Entertainment Law Journal -- thank you for such a great partnership on this!

    New from the inimitable and I: a stab at conceptualizing the parade of horribles that are [insert X] recognition/analysis technologies involving the human body. We term this cluster of applications : 1/n
    Show this thread
    3 replies 13 retweets 62 likes
    Undo
  19. Oct 28

    Dynamic pricing: Unexpected edge cases, ethics, a few minutes on the "Disparate Impact of Artificial Intelligence Bias in Ridehailing Economy's Price Discrimination Algorithms" regulation, digital cartels, law, and privacy and fairness in machine learning

    In a new episode of , we explore what happens when a machine determines the price you pay.
    1 retweet 2 likes
    Undo
  20. Retweeted
    Oct 22

    Please spread the word— is seeking applications for the CITP Fellows Program for 2022-23.

    28 retweets 29 likes
    Undo
  21. Retweeted
    Oct 19

    👇I am looking for a postdoc. Know someone with interest or experience in online mis(dis)information/influence operations, competent in social media analytics & wants to work in a non-western context, like the Indian subcontinent. You can ping me directly

    The second postdoc position will work on a project with assistant professor and faculty member that will study how adversarial online influence looks like in the Indian Ocean Region ... Details: (3/3)
    Show this thread
    1 reply 42 retweets 48 likes
    Undo

@aylin_cim hasn't Tweeted yet.

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.

    You may also like

    ·