1. X
  2. Joshua Levy
Log inSign up
Joshua Levy
7,716 posts
Image
user avatar
Joshua Levy
@ojoshe
Words, code, truth, kindness, other arcana ∙ github.com/jlevy
SF
Joined December 2013
5,347
Following
4,974
Followers
RepliesRepliesArticlesArticlesMediaMedia

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up
  • Pinned
    user avatar
    Joshua Levy
    @ojoshe
    Jan 23, 2023
    Replying to @BrianFeroldi and @GRDecter
    The danger is not in letting students use tools like ChatGPT. It is in letting them learn that they don’t have to think, or believing that the tools are smarter than they are.
    user avatar
    Joshua Levy
    @ojoshe
    Jan 23, 2023
    Now tools that create plausible yet dubious content are ubiquitous, good judgement and critical thinking are more important than ever. I’d like to see teachers help students critique AI-written essays. Find errors, describe what is incorrect or what you would write differently.
    203K
  • user avatar
    Joshua Levy
    @ojoshe
    Aug 30, 2025
    Replying to @mickeyxfriedman
    I remember Yahoo Answers. While I agree one has limited life experience at 9 years old, it’s entirely possible you still had the best answers on the site.
    67K
  • user avatar
    Joshua Levy
    @ojoshe
    Nov 20, 2023
    Replying to @RichDecibels
    If I were to give advice to a friend on depression, based on my own experience and that of others I've been close to, it would be this. (I tried to be brief but it's 3 pages.)
    Image
    Image
    Image
    54K
  • user avatar
    Joshua Levy
    @ojoshe
    Jan 3, 2023
    The area of emerging research around sunlight/infrared on COVID outcomes seems so significant, widely applicable, and low risk that far more people should know about it. I don't post about COVID much, but since I've seen little on Twitter about this, a 🧵 may be helpful. 1/13
    56K
  • user avatar
    Joshua Levy
    @ojoshe
    Oct 15, 2025
    Replying to @shwedotai and @marvinliao
    Getting what you want may be a test of intelligence. But realizing you didn’t in fact want what you thought you wanted is a test of wisdom.
    17K
  • user avatar
    Joshua Levy
    @ojoshe
    Jan 13, 2023
    Replying to @AliAbdaal
    I spent 2+ years during the pandemic learning and experimenting with lots of bodyweight/calisthenics fitness programs. Many are great but few give a full picture. So I summarized it all in one chart. It's big. Still a draft. But it's the reference I wish I'd had starting out.
    Image
    33K
  • user avatar
    Joshua Levy
    @ojoshe
    Aug 9, 2025
    Replying to @elenanisonoff
    My new theory is that contrarian almost-crazy-but-could-be-true theory building has always been a strength of startup founders. But that same characteristic makes you highly susceptible to LLM-induced delusions. In fact, it’s like some models hunt them out for you.
    8.6K
  • user avatar
    Joshua Levy
    @ojoshe
    Mar 16, 2025
    Depending on who I talk to, I feel either early or late to the party. But if you build with Python and haven't switched already, @astral_sh uv is awesome. Go for it. A quick 🧵 with a few notes and my own project template, in case others getting on the uv train find it helpful.
    With apologies to the original @xkcdComic https://xkcd.com/1987/
    49K
  • user avatar
    Joshua Levy
    @ojoshe
    Oct 23, 2024
    Replying to @alexalbert__
    This is very cool but I have to ask. If it's so new and different, couldn't @AnthropicAI come up with a better name than "the new Claude 3.5 Sonnet"? Like (just spitballing here) maybe "Claude 3.6"?
    17K
  • user avatar
    Joshua Levy
    @ojoshe
    Mar 14, 2023
    GPT-4 passing the LSAT or GRE is incredibly impressive. At the same time I think we need a reminder of a logical fallacy we’ll see a lot of this week: That software can pass a test designed for humans does not imply it has the same abilities as humans who pass the same test.
    user avatar
    Jim Fan
    @DrJimFan
    Mar 14, 2023
    I don't give a damn about what is or isn't AGI. It doesn't matter. Below is GPT-4's performance on many standardized exams: BAR, LSAT, GRE, AP, etc. The truth is, GPT-4 can apply to Stanford as a student now. AI's reasoning ability is OFF THE CHARTS. Exponential growth is the
    Image
    40K
  • user avatar
    Joshua Levy
    @ojoshe
    Aug 30, 2025
    It is a beautiful font. But in a terminal, my heart still belongs to B612 Mono (with a few modifications to fix its legacy issues).
    Image
    28K
  • user avatar
    Joshua Levy
    @ojoshe
    May 21, 2020
    While packing books today, found one bit of tech history I’m proud to have inherited: The Internet, circa 1982. With the name and phone number of every... single... person.
    Image
    Image
    Image
  • user avatar
    Joshua Levy
    @ojoshe
    Jan 11, 2023
    Replying to @dgardner
    It’s worth remembering it’s not websites that killed most local newspapers. It was killing revenue models, including bundled classifieds. And it wasn’t feed algorithms that caused Facebook to spread misinformation and divisiveness. It was their incentives to drive ad growth.
    4.5K
  • user avatar
    Joshua Levy
    @ojoshe
    Nov 18, 2023
    Replying to @nickcammarata
    Salvational ideology is a hell of a drug. It would be fascinating to travel back in time a century or more to chat with idealistic political radicals (like Marxists during their early years) and compare it to today.
    25K
Advertisement
Advertisement