Rob Miles

@robertskmiles

Educational YouTube, AI/ML, financial trading systems, unfinished games. Music, movies, microcode, and high-speed pizza delivery

London, Nottingham, SF
Vrijeme pridruživanja: travanj 2010.

Tweetovi

Blokirali ste korisnika/cu @robertskmiles

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

  1. Prikvačeni tweet
    28. velj 2017.

    I'm starting my own YouTube channel about AI Safety! Tell your friends if you have some!

    Poništi
  2. 31. sij

    New Podcast Episode: Alignment Newsletter #84: Reviewing AI alignment work in 2018-19

    Poništi
  3. 22. sij

    New Podcast Episode: Alignment Newsletter #83: Sample efficient deep learning with ReMixMatch

    Poništi
  4. proslijedio/la je Tweet
    20. sij

    Dril is as close to a real-life superhero as we can get - Secret Identity, was revealed once but kept secret by mass consensus - Inspirational cultural figure - currently battling against a robot version of himself designed to duplicate and surpass him

    Poništi
  5. 16. sij

    We need to either change english or change the keyboard layout, because having 'provably' and 'probably' so close together is dangerous

    Poništi
  6. 15. sij

    New Podcast Episode: Alignment Newsletter #82: How OpenAI Five distributed their training computation

    Poništi
  7. proslijedio/la je Tweet

    Since Aaron is gone, it's up to the rest of us to embody his principles. Be curious and kind. Help others. "What is the most important thing you could be working on in the world right now? and if you're not working on that, why aren't you?" RIP, Aaron. Our work is never over.

    Poništi
  8. 8. sij

    New Podcast Episode: Alignment Newsletter #81: Universality as a potential solution to conceptual difficulties in intent alignment

    Poništi
  9. 7. sij

    Got an iffy-looking email about 'advertising on my channel', googled it and yeah it's a scam to get you to dl a virus. They'd CCd dozens of youtubers (not BCC!) so I emailed a quick warning to everyone. Now a bunch of us are chatting, checking out eachother's channels etc :)

    Poništi
  10. 4. sij

    Sometimes I get a comment on a video that I want to reply to with: "That's an important point, and if you're interested in that line of thought then boy do I have a rest-of-the-video-you're-currently-watching for you"

    Poništi
  11. 2. sij

    New Podcast Episode: Alignment Newsletter #80: Why AI risk might be solved without additional intervention from longtermists

    Poništi
  12. 1. sij

    "It's Japanese food, so it's healthy"

    Poništi
  13. 1. sij

    New Podcast Episode: Alignment Newsletter #79: Recursive reward modeling as an alignment technique integrated with deep RL

    Poništi
  14. proslijedio/la je Tweet
    1. sij

    ░░░░░░░░░░░░░░░ 0%

    Poništi
  15. 30. pro 2019.
    Poništi
  16. 29. pro 2019.

    There's a narrative that it was a panic over nothing. Similarly, in the best plausible scenario, we solve AI Safety and years later we're remembered as 'silly people concerned about a non-issue' rather than 'sensible people whose concern helped prevent an issue'

    Prikaži ovu nit
    Poništi
  17. 29. pro 2019.

    It recently came to my attention that many otherwise smart and well informed people don't know that the Y2K "Millennium Bug" was a real thing, which would have caused *huge* problems if people hadn't worked hard to fix it in time

    Prikaži ovu nit
    Poništi
  18. 25. pro 2019.

    New Podcast Episode: Alignment Newsletter #78: Formalizing power and instrumental convergence, and the end-of-year AI safety charity comparison

    Poništi
  19. 19. pro 2019.

    I might delete this tweet since it reads like I'm talking about AI in games, rather than what I mean, which is games like Chess, Go, Poker, etc

    Prikaži ovu nit
    Poništi
  20. 19. pro 2019.

    New Podcast Episode: Alignment Newsletter #77: Double descent: a unification of statistical theory and modern ML practice

    Poništi
  21. 19. pro 2019.

    I wonder if people realise how many games have no superhuman AI yet purely because it would cost money to train and you wouldn't learn anything new by doing so, so researchers don't care

    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:

    ·