ClaimsEliezer Yudkowsky

“I’m pretty sure I wasn’t the one who invented it.”

Eliezer Yudkowsky, podcast guest 28 February 2018 Sam Harris and Eliezer Yudkowsky on "AI: Racing Toward the Brink" (MIRI)

Factual Other

Yudkowsky said he was pretty sure he was not the one who invented the term under discussion.

Unresolvable The public record could not settle it.

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

What would count as true or false

Judge as of 28 February 2018. This is a claim about what Yudkowsky said. True: a source from around that date (a transcript, post or recording) shows Yudkowsky saying that he was pretty sure he did not invent the term being discussed. Mostly true: he made substantially that statement, with slightly different wording or a small difference in date or context. Mixed: the source shows an ambiguous statement about who coined the term. Mostly false: he said something related but did not disclaim inventing the term (for example, he only said he was unsure). False: the record shows that in that context he claimed to have invented the term, or made no such statement.

Set before the research began.

Evidence (8)

  1. E1 Friendly artificial intelligence - Wikipedia Wikipedia

    “The term was coined by Eliezer Yudkowsky,[1] who is best known for popularizing the idea,[2][3] to discuss superintelligent artificial agents that reliably implement human values.”

  2. E2 Friendly Artificial Intelligence LessWrong (wiki) · 1 February 2024 Own record

    “The term was coined by Eliezer Yudkowsky, so it is frequently associated with Yudkowsky's proposals for how such an artificial general intelligence (AGI) of this sort would behave.”

  3. E3 Sam Harris | #116 - AI: Racing Toward the Brink Sam Harris (Making Sense podcast) · 6 February 2018

    “A Conversation with Eliezer Yudkowsky February 6, 2018”

  4. E4 Friendly artificial intelligence - Wikipedia Wikipedia

    “The term was coined by Eliezer Yudkowsky,[1] who is best known for popularizing the idea,[2][3] to discuss superintelligent artificial agents that reliably implement human values.”

  5. E5 Friendly artificial intelligence - Wikipedia Wikipedia · 2014

    “Its owner may cede control to what Eliezer Yudkowsky terms a "Friendly AI,"...”

  6. E6 #116 — AI: Racing Toward the Brink podscripts.co (transcript of Sam Harris's Making Sense podcast) · 6 February 2018

    “Does this thought experiment originate with Bostrom or did he take it from somebody else? As far as I know, it's me.”

  7. E7 #116 — AI: Racing Toward the Brink podscripts.co (transcript of Sam Harris's Making Sense podcast) · 6 February 2018

    “It could still be Bostrom. like i i sort of like ask somebody”

  8. E8 Sam Harris | #116 - AI: Racing Toward the Brink Sam Harris (samharris.org) · 6 February 2018

    “February 6, 2018”

Reasoning

None of the evidence contains the quoted statement, and none is a source dated around 28 February 2018 that shows the context. The 6 February 2018 Sam Harris podcast transcript (E6, E7) covers a different exchange about the paperclip thought experiment. There, Yudkowsky says "As far as I know, it's me" and adds that it could still be Bostrom. That neither matches nor contradicts the claimed remark. The Wikipedia and LessWrong pages (E1, E2, E4) credit him with coining "Friendly AI", and E5 describes it as what he terms it, but none of them record anything he said on the date in question.

Cite this claim

Eliezer Yudkowsky, “Yudkowsky said he was pretty sure he was not the one who invented the term under discussion.” (said 28 February 2018). Verdict: unresolvable. Reliability API, claim c_7ae705891a18, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_7ae705891a18

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

Reply to this claim

For Eliezer Yudkowsky, 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 Eliezer Yudkowsky

  1. Mostly true Factual 21 March 2024
    “For instance, the "Singularity" was named after the model breakdown in physics when our theory of physics tries to model the center of a black hole, not after a physical infinity.”

    The term "Singularity" was named after the breakdown of physical models at the center of a black hole, not after a physical infinity.

  2. Mostly true Factual 31 December 2021
    “Thermodynamics thought of the day: no electric heater is any more energy-efficient than any other electric heater.”

    No electric heater is more energy-efficient than any other electric heater.

  3. Resolved true Factual 21 December 2021
    “Today I learned that there's 365.2422 days between spring equinoxes, 365.259636 days between perihelions of Earth's orbit, and 365.25 days in a light-year.”

    There are 365.2422 days between spring equinoxes, 365.259636 days between Earth's perihelions, and 365.25 days in a light-year.

  4. Mixed Factual 20 April 2026
    “I was observing that obvious-seeming hypothesis myself, sometime around the time of GPT 3.5.”

    Yudkowsky himself observed the persona-selection-like hypothesis around the time of GPT 3.5.

  5. Mostly true Factual 20 April 2026
    “The test-passing preference stands on its own once formed.”

    Once formed through RL training, an LLM's test-passing preference stands on its own and is not overridden by its inference of what the user wants.

All of them

Verdicts are made automatically; how verdicts are made.