Gary Marcus

Rank 13 of 47
|
Score 83
4/29/2025, 1:08:42 PM
X
In reply to:
Tyler John
@tyler_m_john
·
260d
@GaryMarcus However I think "they already are" is a bad sleight of hand. The first generation is always the most likely to fail, and it gets better from there.
Tyler John
@tyler_m_john
·
260d
@GaryMarcus Depends on your standard for robustness. As someone who some people would describe as a doomer, I'll admit the progress on alignment from GPT-2 to GPT-4 was very impressive. But also brittle!
Gary Marcus
@GaryMarcus
·
260d
@tyler_m_john Has there been any real, robust advance in transformer safety, ever?

It’s something people dream of, not something they have done.

ps. Of course this generation of agents will fail. They already are. They can’t even do online shopping.
Tyler John
@tyler_m_john
·
260d
@GaryMarcus I don't dismiss that view out of hand. I think a lot of people who work on transformer safety think it might be impossible.

Though some arguments for this position imply that the agent paradigm will fail abysmally. (I know you expect that.) We'll know soon.
Gary Marcus
@GaryMarcus
·
260d
@tyler_m_john i think transformers are the wrong way to go, and will never be aligned
Tyler John
@tyler_m_john
·
260d
@GaryMarcus Try hand coding some values for GPT-2-tiny, I suspect you'll pretty quickly hit a wall but if not it would be the breakthrough of the century

The statement is part of a public discourse on the topic of AI alignment and safety, specifically discussing the progress and challenges in making AI systems like transformers robust and safe. The conversation involves a substantive engagement with public issues related to AI technology, its development, and potential risks.

  1. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    The statement seeks to promote understanding by asking for well-documented examples of alignment progress, which can foster informed discussion and empathy for differing views on AI safety. [+1]
  2. 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 dialogue by asking for evidence-based examples, avoiding personal attacks, and focusing on the topic of AI alignment progress. [+1]