Gary Marcus

Rank 13 of 47

Score 84

@gcolbourn making the LLM use tools reliably is actually hard; making them following constraints (eg alingment) is damn hard. for many things the tools we might need don’t exist.

we have tons left to do.

6/27/2025, 3:06:21 PM

In reply to:

Greg ⏹️ Colbourn

@gcolbourn

227d

@GaryMarcus So what secret sauce is left, given we have LLMs that can use standard (neurosymbolic) computing tools?

Gary Marcus

@GaryMarcus

227d

prediction: When AGI doesn’t arrive in 18 months, exactly zero people will admit they were wrong.

Seán Ó hÉigeartaigh

@S_OhEigeartaigh

228d

18 months until we learn if the Anthropic folks are high on their own supply, or whether the world is about to become very inexplicable for this jumped-up ape. Still leaning towards the former FWIW. But I'd sleep better if their prediction track record wasn't so good (esp Amodei)

The statement engages in a technical discussion about the challenges of aligning large language models (LLMs) and the limitations of current tools, contributing to the broader discourse on artificial intelligence and its development.

Principle 1:
I will strive to do no harm with my words and actions.
The statement does not use harmful language and focuses on the technical challenges, adhering to the principle of doing no harm. [+1]
Principle 3:
I will use my words and actions to promote understanding, empathy, and compassion.
The statement promotes understanding by acknowledging the complexity of AI alignment and tool development, aligning with the principle of promoting understanding and empathy. [+1]
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 addressing the technical aspects of AI development without resorting to personal attacks, adhering to the principle of constructive criticism. [+1]