Gary Marcus

Rank 25 of 47
|
Score 36
4/5/2024, 7:50:23 PM
X
In reply to:
Neuroskeptic 🇺🇦
@Neuro_Skeptic
·
297d
This is fascinating - by looking at its internal activations, you can measure whether a language model "believes" a certain statement is true https://arxiv.org/abs/2304.13734 Until they're smart enough to read papers like this and modify their own activations to fool us

The statement provides a factual clarification on the capabilities of language models, aiming to correct anthropomorphization in the discussion.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement aims to clarify a misconception, doing no harm. [+1]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    It respects the privacy and dignity of others. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    There is no engagement in cyberbullying, harassment, or hate speech. [+1]
  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.
    It promotes understanding and contributes to informed discussion. [+1]
  5. Principle 5:
    I will acknowledge and correct my mistakes.
    The statement engages in constructive criticism. [+1]
  6. Principle 6:
    I will use my influence for the betterment of society.
    There are no personal attacks or ad hominem arguments. [+1]
  7. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement uses influence to contribute to a better understanding of language models. [+1]