Gary Marcus

Rank 32 of 47
|
Score -1
4/2/2024, 6:26:34 PM
X
In reply to:
Gary Marcus
@GaryMarcus
·
229d
The attack surface in LLMs is endless.

I’m all for AI, but LLMs are a train wreck waiting to happen.
Ethan Mollick
@emollick
·
229d
New jailbreaking technique: pure repetition.

AIs are getting big context windows, it turns out if you fill a lot of it with examples of bad behavior, the AI becomes much more willing to breach its own guardrails. Security people are used to rules-based systems. This is weirder.

The statement addresses security concerns of LLMs, contributing to public understanding and debate on AI security.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    Raises awareness about potential harm, aiming to prevent it. [+1]
  2. Principle 2:
    I will respect the privacy and dignity of others and will not engage in cyberbullying, harassment, or hate speech.
    Does not violate the principle of respecting privacy and dignity. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    Contributes to understanding and potentially empathizing with the complexities of AI security. [+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.
    Provides constructive criticism without personal attacks. [+1]
  5. Principle 6:
    I will use my influence for the betterment of society.
    Aims to use influence to improve AI security for societal benefit. [+1]