Karen HaoOvjeren akaunt

@_KarenHao

demystifying AI . formerly: data scientist & reporter , engineer. sign up for my 💌: .

Cambridge, MA
Vrijeme pridruživanja: veljača 2016.

Tweetovi

Blokirali ste korisnika/cu @_KarenHao

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @_KarenHao

  1. Prikvačeni tweet

    IT'S HERE!!! The biggest story I’ve ever worked on. 's very first interactive ever, which walks through a concrete example of AI bias, and why it’s so much more complicated than initially meets the eye. 1/

    Prikaži ovu nit
    Poništi
  2. Wow this really punches you in the gut

    Poništi
  3. Haha my expression and giant hand. Thank you for having me!!

    Poništi
  4. 3. velj

    Me and my editor on the reg 😂

    Poništi
  5. 1. velj

    Natural language processing has seen some stunning advancements in the last two years. But a new study shows there’s still something missing.

    Poništi
  6. 31. sij
    Poništi
  7. 30. sij

    Also so many photos of women!! A wonderful reminder of the foundational role women played in the early days of computing.

    Prikaži ovu nit
    Poništi
  8. 30. sij

    I get a ton of unsolicited books sent to me all the time, and every once in a while, it’s a true gem. I’m so excited about this one from presenting a visual history of punch cards. So many gorgeous phots from a time when data was physical. 😍

    Prikaži ovu nit
    Poništi
  9. 29. sij

    Research labs have made incredible advances in combining AI & robotics, but bringing them into the real-world has been a completely different story. I visited that says it has cracked the nut and is ready to take on warehouse floors by storm.

    Poništi
  10. 27. sij

    So I'm doing a thing next month. 👀 If you're in Mumbai, come through! 🤗

    Poništi
  11. 25. sij

    Happy Lunar New Year everyone!! 🤗 May the dumplings be ever in your flavor 🥟🥟

    Poništi
  12. 22. sij
    Poništi
  13. 19. sij

    These thoughts are also still nascent, so any feedback welcome! And now I will stop procrastinating on the other work I need to do. 🤗 12/

    Prikaži ovu nit
    Poništi
  14. 19. sij

    Of course there are some important differences between nuclear & AI. For one, the barriers to access are much higher for nuclear, much less for AI. But there are still important lessons to learn from this historical parallel. 11/

    Prikaži ovu nit
    Poništi
  15. 19. sij

    It was the first major step to nuclear weapons control and laid an important foundation for the international Non-Proliferation Treaty (NPT), which was negotiated b/w 1965 and 1968 and entered into effect in 1970. 10/

    Prikaži ovu nit
    Poništi
  16. 19. sij

    It wasn't until the US moved from attacking physics to more concrete regulation of nuclear weapons that things actually started to get somewhere. In 1963, the US, Soviet Union & Great Britain signed the Limited Nuclear Test Ban Treaty, restricting testing of nuclear weapons. 9/

    Prikaži ovu nit
    Poništi
  17. 19. sij

    So it's much more sensible to talk about export controls at the model or application level, eg on face recognition systems. Not only is this more feasible, history also tells us it will be more effective! 8/

    Prikaži ovu nit
    Poništi
  18. 19. sij

    It was really ineffective, humiliated a lot of great scientists including Oppenheimer, & led to a dramatic drop in physics enrollment at universities. Basically, it was just anti-science & left a black mark on US history. (Shout out to my history of science prof David Kaiser.) 7/

    Prikaži ovu nit
    Poništi
  19. 19. sij

    Trying to regulate AI at this level, eg by export controlling algorithms, makes no sense. It's like when the US gov went after physics & physicists during the Red Scare in the 40s/50s out of a fear that adversaries would get hold of "nuclear secrets" & build nuclear weapons. 6/

    Prikaži ovu nit
    Poništi
  20. 19. sij

    When people talk about the open-source nature of AI, they're mostly talking about the algorithm level. Fundamental algorithm advancements have always happened with significant collaboration across borders. Here's an example of why this is a good thing! 5/

    Prikaži ovu nit
    Poništi
  21. 19. sij

    Third is the application layer: the model or set of models applied to a task within a specific context. 4/

    Prikaži ovu nit
    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·