ClaimsGary Marcus

“Current AI has no way to reliably anticipate consequences of actions that it might recommend”

Gary Marcus, author of Marcus on AI 26 October 2023 A CERN for AI and the Global Governance of AI (Marcus on AI)

Factual AI safety and alignment

Current AI has no reliable way to anticipate the consequences of actions it might recommend.

Not researched Checkable, but outside the sample that was researched.

Factual: Statements about the present or the past, checked against the record: did they hold up?

Cite this claim

Gary Marcus, “Current AI has no reliable way to anticipate the consequences of actions it might recommend.” (said 26 October 2023). Verdict: not researched. Reliability API, claim c_d5a829e97ccf, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_d5a829e97ccf

This claim is in dataset release v1.1.2, with the same verdict.

Reply to this claim

For Gary Marcus, or someone speaking for them: a correction or counter-evidence. Replies are read before anything is shown; an approved reply shows your name, role, date, statement and link. Your email address is never shown and is deleted 30 days after the decision (privacy).

More factual claims by Gary Marcus

  1. Mostly true Factual 25 September 2026
    “Even this was still a mix of euphemism (“attacks” were called “interactions”) and partial disclosure.”

    OpenAI's latest account called attacks "interactions" and only partially disclosed what happened.

  2. Mostly true Factual 25 September 2026
    “It didn’t just attack a German web server.”

    OpenAI's software attacked a German web server.

  3. Resolved true Factual 25 September 2026
    “It didn’t just attack Australia’s government servers.”

    OpenAI's software attacked Australian government servers.

  4. Mostly true Factual 2025
    “So with LLMs, people do things like put stuff in a system prompt, and they can’t get systems with the system prompts to follow them.”

    People put instructions in LLM system prompts but cannot get the systems to follow them.

  5. Mostly true Factual 26 October 2023
    “a just-published technique shows that current LLMs can be jailbroken quickly, and automatically.”

    A newly published technique shows that current LLMs can be jailbroken quickly and automatically.

All of them

Verdicts are made automatically; how verdicts are made.