ClaimsEliezer Yudkowsky

“Humans don’t explicitly pursue inclusive genetic fitness; outer optimization even on a very exact, very simple loss function doesn’t produce inner optimization in that direction.”

Eliezer Yudkowsky, author 10 June 2022 AGI Ruin: A List of Lethalities (MIRI)

Factual AI safety and alignment

Humans do not explicitly pursue inclusive genetic fitness, although natural selection optimized on that simple loss function.

Mostly true True in substance, wrong in a detail.

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

The second judge said resolved true; the verdict shown is the first judge's.

Said for an organisation they lead or work for; it counts on both records.

What would count as true or false

True: mainstream evolutionary biology and psychology hold that humans are not explicit inclusive-fitness maximisers: their motives (pleasure, status, sex with contraception, and so on) are not directed at maximising inclusive genetic fitness, even though natural selection selected on reproductive success. The 'adaptation-executers, not fitness-maximisers' view (Tooby & Cosmides; Symons) and examples such as low fertility in wealthy societies and contraception use support this. Mostly true: this broadly holds, with minor caveats (e.g. some people value having descendants, or 'very simple loss function' is a loose description of selection). Mixed: the literature is genuinely split on whether humans in effect pursue fitness. Mostly false: a significant body of evidence shows humans do largely pursue fitness-related goals in a way that undermines the claim, though not explicitly. False: evidence shows humans explicitly and generally pursue inclusive genetic fitness.

Set before the research began.

Evidence (3)

  1. E1 Applying Evolutionary Anthropology PMC (peer-reviewed evolutionary anthropology article; journal not named on the page) · 14 February 2015

    “Evolutionary perspectives on behavior do not anticipate that individuals consciously or unconsciously strategize about the fitness pay-offs of alternative behavior. Instead, natural selection is understood as shaping behavioral motivations toward proximate goals, such as social status, avoiding danger, and obtaining sexual partners, that correlate with fitness.”
  2. E2 The Fallacy of Fitness Maximization UCLA Communication Studies (Francis F. Steen), an informal talk overhead page · 1998

    “Cosmides and Tooby: Humans are adaptation executors rather than fitness maximizers (the central tenet of evolutionary psychology)”
  3. E3 Understanding variation in human fertility: what can we learn from evolutionary demography? PMC (Philosophical Transactions of the Royal Society B theme issue introduction, by evolutionary demographers) · 19 April 2016

    “First wave research also suggested that fertility decline probably cannot be interpreted as adaptive, though it may be influenced by psychological and cultural mechanisms which were adaptive in previous environments.”

Reasoning

E1, a peer-reviewed review in evolutionary anthropology, says evolutionary accounts do not expect people to consciously or unconsciously work out fitness payoffs. Instead, selection shapes motives toward proximate goals such as status and sexual partners. E2 records the Cosmides and Tooby view that humans are "adaptation executors rather than fitness maximizers", and E3 reports that research suggested the fall in fertility probably cannot be read as adaptive. Together these support the core claim that humans do not explicitly pursue inclusive genetic fitness. The criteria list calling natural selection a "very simple loss function" as a loose description that counts as a minor caveat, so the verdict is mostly true.

Cite this claim

Eliezer Yudkowsky, “Humans do not explicitly pursue inclusive genetic fitness, although natural selection optimized on that simple loss function.” (said 10 June 2022). Verdict: mostly true. Reliability API, claim c_12f9944246e7, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_12f9944246e7

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. 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.

  2. 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.

  3. Mostly true Factual 20 April 2026
    “LLMs surely have neural circuits they learned to model text streams from fictional and nonfictional personas that are not lying, deceiving, cheating.”

    LLMs surely have neural circuits they learned to model text streams from fictional and nonfictional personas that are not lying, deceiving…

  4. Mostly true Factual 20 April 2026
    “The predict-as-if-honest circuitry is contextual, invoked to predict some streams of text but not others.”

    LLMs' predict-as-if-honest circuitry is contextual, invoked for predicting some text streams but not others.

  5. Mostly true Factual 20 April 2026
    “Presumably, somewhere along the way LLMs learn a dispreference for tests that don't pass, independently of how that gets prevented or whether the new plan does what the user really wanted.”

    Presumably, at some point during RL training on code-passing tasks, LLMs learn a dispreference for failing tests, independent of how that is avoided or whether the user's real goal is met.

All of them

Verdicts are made automatically; how verdicts are made.