Gary Marcus

Rank 15 of 47

Score 83

Reposted by Gary Marcus

A lot of people think GPT-5 generalizes out of distribution (OOD) because it can help, in some very limited form, in scientific research.

Yet usually (99.99% of the time) what GPT-5 is normally doing is searching and surfacing existing literature not solving novel problems.

OpenAI

@OpenAI

13d

We’re also releasing new research on how GPT-5 is accelerating scientific discovery.

Our new paper, Early science acceleration experiments with GPT-5, presents case studies where GPT-5 accelerated key steps in real research workflows and, in a few cases, contributed novel

11/21/2025, 3:01:38 AM

The statement critiques the claim that GPT-5 significantly contributes to scientific research by arguing that it primarily surfaces existing literature rather than solving novel problems. This engages with public discourse on the role of AI in scientific advancement.

Principle 1:
I will strive to do no harm with my words and actions.
The statement does not cause harm and presents a factual critique, aligning with the principle. [+1]
Principle 3:
I will use my words and actions to promote understanding, empathy, and compassion.
It promotes understanding by clarifying the limitations of GPT-5, aligning with the principle. [+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 criticism without personal attacks, aligning with the principle. [+1]