Log inSign up
Adrien Ecoffet
3,953 posts
@AdrienLE

Adrien Ecoffet

@AdrienLE
Trying to make AGI go well. Researcher at @openai. Views my own.
San Francisco, CA
adrien.ecoffet.com
Joined April 2009
235
Following
8,223
Followers
RepliesRepliesRepostsRepostsMediaMedia

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
  • @AdrienLE
    Adrien Ecoffet
    @AdrienLE
    Aug 11
    Whose picture would @nypost use if they wrote a hit piece about you?
    16
  • @AdrienLE
    Adrien Ecoffet
    @AdrienLE
    Aug 11
    The Dean Ball meme template would have been so good for when Anthropic hired Kyle Fish (below)
    Image
    1
  • @AdrienLE
    Adrien Ecoffet
    @AdrienLE
    Aug 7
    If the historical LLM critics are now claiming that LLM + harness = neurosymbolic AIs, it must mean either that: 1/ they thought the LLM cos would never implement harnesses (wrong) 2/ they think the LLM cos only ended up implementing harnesses because of their warnings (absurd)
    1
  • @AdrienLE
    Adrien Ecoffet
    @AdrienLE
    Aug 4
    >friend tells me about a new evil AI company >ask her if it’s evil or just successful >she doesn’t understand >pull out diagram about how when AI companies become successful, arrows start flying at them out of nowhere until a competitor catches up >she laughs and says “it’s a
    3
  • @AdrienLE
    Adrien Ecoffet
    @AdrienLE
    Jul 29
    The one thing that the entire AI ecosystem can agree on, from Pause to e/acc, is that Diet Coke is good, actually
    @daniel_271828
    zzDANIEL_ETH
    @daniel_271828
    Jul 29
    These compounds all work by different mechanisms, right? And IIUC, some of them (eg aspartame) it would be mechanistically surprising if they caused long-term neurological decline. This leads me to believe the association isn’t causal. The one common factor is the consumer.
    2
Advertisement
Advertisement