Pundit ClaimsAI claims by Gary Marcus and Ed Zitron, with the evidence

Gary Marcus Marcus on AI Published

Stop Treating AI Models Like People

We recorded 9 claims from this post. 3 were checked against evidence. 6 can’t be checked. 3 more are held. Read the post at its source.

Claims from this post (9)

In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.

  1. Assessment: SupportedSaid

    “As most experts realize, the reality is that current AI doesn’t “decide to teach itself”, or even have consistent beliefs.”

    What we tested: Current AI does not decide to teach itself and does not have consistent beliefs.

    What happened: The report describes externally organized training and explicitly states in its limitations that GPT-4 does not learn from its experience. This supports the operational reading of the self-teaching clause. (OpenAI, )

  2. Assessment: Can’t be checkedSaid

    “The AI systems that power these chatbots are simply systems (technically known as “language models” because they emulate (model) the statistical structure of language) that compute probabilities of word sequences, without any deep or human-like comprehension of what they say.”

    What we tested: Language models powering chatbots compute word-sequence probabilities without deep or human-like comprehension of what they say.

  3. Assessment: UnresolvedSaid

    “With technology more and more able to emulate human responses based on larger and larger samples of text (and “reinforcement learning” from humans who instruct the machines), the problem has grown even more pernicious.”

    What we tested: Growing ability of AI to emulate human responses has made the overattribution (ELIZA effect) problem more pernicious.

  4. Assessment: UnresolvedSaid

    “When we do so, we amplify the hype around AI, and lead people into thinking that these machines are trustworthy oracles capable of manipulation or decision-making, which they are not.”

    What we tested: When we treat AI models like people, we amplify the hype around AI and lead people into thinking these machines are trustworthy oracles capable of manipulation or decision-making, which they are not.

  5. Assessment: Can’t be checkedSaid

    “The more false agency people ascribe to them, the more they can be exploited, suckered in by harmful applications like catfishing and fraud, as well as more subtly harmful applications like chatbot-assisted therapy or flawed financial advice.”

    What we tested: The more false agency people ascribe to chatbots, the more they can be exploited via fraud, catfishing, chatbot therapy or flawed financial advice.

  6. Assessment: Can’t be checkedSaid

    “We also need new technical tools, like watermarks and generated content detectors, to help distinguish human- and machine-generated content, and policy measures to limit how and where AI models can be used.”

    What we tested: New tools like watermarks and detectors, plus policy measures limiting how and where AI models can be used, are needed.

  7. Assessment: Can’t be checkedSaid

    “As tempting as it is, we have to stop treating AI models like people.”

    What we tested: People must stop treating AI models like people.

  8. Assessment: Can’t be checkedSaid

    “Educating people to overcome the overattribution bias will be a vital step; we can’t have senators and members of the AI community making the problem worse.”

    What we tested: Educating people to overcome overattribution bias is vital, and senators and AI community members should not worsen it.

  9. Assessment: Can’t be checkedSaid

    “Treat them as fun toys, if you like, but don’t treat them as friends.”

    What we tested: People should not treat chatbots as friends.

Held from this post (3)

These records did not pass every check, so nothing about them is published. Held is not a finding.