Gary Marcus

Rank 26 of 47
|
Score 54
7/15/2024, 5:59:58 PM
X
In reply to:
Jai Bhagat
@jkbhagatio
·
101d
@GaryMarcus Curious on your thoughts on Mechanistic Interpretability for LLMs? E.g. Anthropic's work here: https://transformer-circuits.pub…
Gary Marcus
@GaryMarcus
·
101d
Don’t panic — I promise you GPT-5 will NOT be the Titanic.

But yes we have no idea how generative AI works inside
and we should stop empowering unreliable systems we don’t understand — and instead focus on developing neurosymbolic approaches that are more interpretable.
ControlAI
@ai_ctrl
·
106d
Ex-OpenAI safety researcher William Saunders:
— We fundamentally don't know how AI works inside
— A lot of people in OpenAI think we could be 3 years away from something dangerous
— GPT-5 could be the Titanic

The statement '@jkbhagatio may write about it next week!' is a response within a broader conversation about the interpretability and safety of large language models (LLMs), specifically referencing mechanistic interpretability and neurosymbolic approaches. This conversation involves public concerns about AI safety and transparency, thus constituting public discourse.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does not cause harm and is neutral in tone.
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    The statement respects privacy and dignity, and does not engage in cyberbullying, harassment, or hate speech.
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement is neutral and does not actively promote understanding, empathy, or compassion.
  4. Principle 4:
    I will engage in constructive criticism and dialogue with those in disagreement and will not engage in personal attacks or ad hominem arguments.
    The statement does not engage in personal attacks or ad hominem arguments, and is part of a constructive dialogue. [+1]
  5. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement upholds the principles of free speech and uses the platform responsibly. [+1]