Log inSign up
Image
user avatar
Joel Lehman
@joelbot3000
ML researcher, co-author Why Greatness Cannot Be Planned. Creative+safe AI, philosophy. SiR @second_natureai; prev OpenAI / Uber AI / Geometric Intelligence
San Francisco, CA
joellehman.com
Joined June 2009
1,975
Following
6,815
Followers
  • Pinned
    user avatar
    Joel Lehman
    @joelbot3000
    Jan 24, 2025
    new paper: "Evolution and the Knightian Blindspot of Machine Learning" Our ever-changing world bubbles with surprise and complexity. General AI must include handling unforeseen situations with grace. Yet this issue largely lies outside AI's formalisms: a blind spot. (1/n)
    77K
  • user avatar
    Joel Lehman
    @joelbot3000
    Jun 20, 2022
    “Evolution through Large Models” – new paper from our team at OpenAI. Step towards evolutionary algos that continually invent and improve at inventing: Large models can suggest (+ improve at making) meaningful mutations to code. Paper: arxiv.org/abs/2206.08896 1/4
    Image
    00:00
  • user avatar
    Joel Lehman
    @joelbot3000
    Feb 23, 2023
    "Machine Love" - new paper w/ LM expts: Leaving to humans what is human (emotions/relationships/autonomy), is there any conception of love fitting for machines to embody? One idea: to provide unconditional support for humans to pursue their own growth. arxiv.org/pdf/2302.09248…
    Image
    00:00
    56K
  • user avatar
    Joel Lehman
    @joelbot3000
    Jan 9, 2019
    Happy to announce the Atari Zoo -- a collection of trained reinforcement learning agents and open-source software to help catalyze their understanding: eng.uber.com/atari-zoo-deep… w/ many great collaborators at UberAI, Google Brain, and OpenAI
  • user avatar
    Joel Lehman
    @joelbot3000
    Sep 9, 2020
    I’m excited to announce that I have joined @OpenAI as of last month, am happy to again team up with the singular @kenneth0stanley to push forward the frontiers of open-endedness. Unexpected where the stepping stones lead (image from 1st academic presentation at ALife 2008). 1/3
    Image
  • user avatar
    Joel Lehman
    @joelbot3000
    Jun 20, 2022
    Replying to @joelbot3000
    And a bonus video of some favorite ELM robots
    Image
    00:00
  • user avatar
    Joel Lehman
    @joelbot3000
    Mar 25, 2024
    Side project: Identifying Life-Changing Books with LLMs tldr: Language models can analyze millions of book reviews to identify books that readers are most likely to say "changed their life." blog (w/ complete list of 300 most 'life-changing' books): flourish.ing/identifying-li…
    21K
  • user avatar
    Joel Lehman
    @joelbot3000
    Sep 9, 2019
    'I hate crowds in science ... I tell my students, when you go to these meetings, see what direction everyone is headed, so you can go in opposite direction. Don't polish the brass on the bandwagon' -VS Ramachandran on big conferences c/o The Brain that Changes Itself
  • user avatar
    Joel Lehman
    @joelbot3000
    Jul 16, 2020
    Happy to announce “Reinforcement Learning Under Moral Uncertainty” paper led by @AdrienLE. We don’t know for sure which theory of morality is correct, but still want agents to act ethically as possible. We bridge moral philosophy and ML towards that goal: arxiv.org/abs/2006.04734
    Image
    00:00
  • user avatar
    Joel Lehman
    @joelbot3000
    Jan 11, 2025
    Replying to @RomeoStevens76
    Hadn't heard of this book before, then did an LLM project to go through goodreads reviews (to rank books by what percentage of reviews mentioned it changed their life) -- and this one came in #2, behind another one I hadn't heard of before blog.joellehman.com/identifying-li…
    19K
  • user avatar
    Joel Lehman
    @joelbot3000
    Jan 12, 2024
    (1) I think AI safety is an important and philosophically deep research area. (2) I think AI safety is likely pre-paradigmatic - that much of what we think we know is likely to be misleading and future research will overturn present formalisms. (3) Given that, and how
    16K
  • user avatar
    Joel Lehman
    @joelbot3000
    Jun 20, 2022
    Replying to @joelbot3000
    Synergy between evolutionary algorithms and language models: EAs can leverage LMs to generate data in novel domain, and thereby improve LM — bootstrapping it into competence in that domain (here evolving locomoting 2D virtual creatures). 3/4
    Image
  • user avatar
    Joel Lehman
    @joelbot3000
    May 22, 2016
    Think occasionally of the suffering of which you spare yourself the sight. --Albert Schweizer
  • user avatar
    Joel Lehman
    @joelbot3000
    Jun 20, 2022
    Replying to @joelbot3000
    ELM done with an amazing team: @kenneth0stanley @GordonJo76 @darkmatter08 @kandouss & Cathy Yeh. [video shows an evolutionary lineage from an initial barely-functional seed program to a final solution -- best in full-screen] 2/4
    Image
    00:00