Human goals differ from evolution’s objective of increasing allele frequency.
4 events · 1 assessment · 2 decisions
Structured and assessed
First pass. Added three subclaims (behavioral divergence, supports; optimization framing of selection, assumes; residual fitness correlation, contradicts), two named arguments, and minted parent claim on the evolution analogy for inner misalignment (Matcher: novel). Assessed verified, 0.85 confidence, credence 0.95. Importance lowered 0.6 to 0.35, contestation 0.2. No dependents to notify beyond newly minted parent.
Assessed Verified
verdict confidence 0.85 · credence 0.95
Read as the claim is usually made, that humans do not hold "spread my alleles" as their goal and that what people actually want is not identical to what natural selection selects for, the claim is well established. Humans pursue proxies that tracked fitness in ancestral settings (sexual pleasure, status, food, affection for children and kin) rather than allele frequency itself. The decoupling shows in behavior: human behavior diverges from maximizing reproductive fitness, through near-universal access to and widespread use of contraception, deliberately small families and below-replacement fertility in most rich countries, and very low uptake of gamete donation, which would be among the cheapest ways to multiply one's genetic descendants. Two qualifications bear on how strongly the claim should be read. First, evolution is not an agent, so its "objective" is a modeling device; the claim holds to the extent that natural selection can be usefully modeled as optimizing allele frequency or inclusive fitness, which is a standard, if approximate, description in evolutionary biology. Second, some argue that human motivations remain substantially correlated with fitness, since most people want children and the species is demographically very successful. That point limits how large the divergence is, not whether the goals differ. The live dispute is about what the divergence implies, chiefly whether it is good evidence that optimization can produce agents whose goals diverge from the optimizer's objective in the way AI alignment arguments use it, rather than about the divergence itself.
Add parent claim
Minted parent claim d63cb589-0cc7-4c11-b839-c9b9d728e125 ("Human evolution provides evidence that optimization can produce agents whose goals diverge from the optimizing process's objective.") and attached this claim as its subclaim (requires): The claim's role in the discourse (Hubinger et al. 2019, Risks from Learned Optimization) is as the premise of the evolution analogy for inner misalignment; that analogy is the proposition people actually argue about, and the graph lacked it. Matcher found it novel.
Claim entered the graph