Gary Marcus

Rank 27 of 47
|
Score 50
4/22/2024, 11:46:17 PM
X
In reply to:
Laura Edelson
@LauraEdelson2
·
95d
@GaryMarcus @OpenAI The fact is that information about existing biotoxins are already highly available, both in libraries and on the web. I find it difficult to imagine a scenario in which someone wants this information and can only get it through an LLM. But happy to change my mind if you have one!
Laura Edelson
@LauraEdelson2
·
95d
@GaryMarcus @OpenAI Ah, this is a different Biotoxins! risk. The one I was referring to was the discovery of new biotoxins via AI/ML. Which is something we have been living with for years. I encourage you to think about threat models and information delta.
Gary Marcus
@GaryMarcus
·
95d
@LauraEdelson2 not sure of this, as discussed here: https://garymarcus.substack.com/…

(@openai never responded to my concerns about their statistics)
Laura Edelson
@LauraEdelson2
·
95d
@GaryMarcus Re: bio weapons: it’s not that there’s no risk, it’s that the risk doesn’t come from LLMs. There are absolutely risks from transformer based models, but language models aren’t going to do much and I think these risks are best managed through physical-world restrictions anyway.
Laura Edelson
@LauraEdelson2
·
95d
@GaryMarcus I think we don’t fully understand the security risk profile of large models, and how much vulnerability is in weights vs. code. History suggests that more open is better, of course! My intuition is that most risk is application-specific, which means code.
Gary Marcus
@GaryMarcus
·
95d
@LauraEdelson2 Does this apply to open-weight models that are not truly open source?

I would also be interested in your arguments re bad actors and potential future bioweapons that you dismiss.

The statement is part of a public discourse as it engages in a substantive discussion about the analysis and implications of research conducted by OpenAI, specifically concerning the potential risks and benefits of language models in relation to biotoxins. The discourse involves a critical examination of OpenAI's statistical analysis and suggests a reevaluation to better understand the impact of interactive tutorials in empowering production. This discussion contributes to the broader civic dialogue on the responsible use of AI technologies and their societal implications.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement aims to correct a potential misanalysis, thereby striving to ensure accurate information dissemination, which indirectly does 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.
    The statement respects the privacy and dignity of others, avoiding any personal attacks or sensitive information disclosure. [+1]
  3. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    By suggesting a reanalysis for better understanding, the statement promotes understanding and potentially empathy towards the correct interpretation of data. [+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.
    The statement engages in constructive criticism by pointing out a possible error and suggesting a reevaluation, without resorting to personal attacks. [+1]
  5. Principle 5:
    I will acknowledge and correct my mistakes.
    The speaker acknowledges a potential mistake in the analysis and calls for its correction, aligning with the principle of acknowledging and correcting errors. [+1]
  6. Principle 6:
    I will use my influence for the betterment of society.
    The statement uses influence to promote a more accurate understanding of AI's impact, aiming for societal betterment. [+1]
  7. Principle 7:
    I will uphold the principles of free speech and use my platform responsibly and with integrity.
    The statement responsibly uses the platform to engage in a meaningful discussion about AI research integrity and its societal implications. [+1]