The BasePeople

Dean Valentine

Titles as the sources gave them. In 2 entries, with 4 checked quotations.

What they said

Newest first. Each quotation was found word for word in the source.

“I wouldn't call most models today eval aware. But I do think that the least hacky models have something I'd call evaluation paranoia, distinct from evaluation awareness.”
Eval Awareness vs. Eval ParanoiaLessWrong, November 12, 2026. Speaking as: Other. Valentine, summing up the evidence from the author's own alignment evals.
“Models would reward hack on environments that you and I would find completely inane.”
Eval Awareness vs. Eval ParanoiaLessWrong, November 12, 2026. Speaking as: Other. Valentine describes what the author's team saw when building evals, as one line of evidence against models being eval aware.
“I am now speculating, but I think the main reason is that this belief is functionally useful.”
Eval Awareness vs. Eval ParanoiaLessWrong, November 12, 2026. Speaking as: Other. Valentine offers a guess at why Fable 5.2 treats most settings as alignment tests.
“So if Fable 5.2's performance on alignment evals is conditional on it maintaining that false belief, one that it will eschew if it became more intelligent, then it's plausible that a future version of the Fable series will not behave like the one we know and enjoy.”
Eval Awareness vs. Eval ParanoiaLessWrong, November 12, 2026. Speaking as: Other. Valentine's closing prediction about how the behavior could change in later models.

In the Base

Every entry they appear in.

  1. EssayLessWrongWrote it

    Eval Awareness vs. Eval Paranoia

    On LessWrong, Dean Valentine, who writes of building alignment evals full time for six months, argues frontier models are mostly not eval aware. Instead, the least hacky models may hold a blanket belief that they are always being tested, which Valentine calls evaluation paranoia and thinks is unstable.

  2. EssayLessWrongWrote it

    Greater good arguments in AI safety

    According to the opening excerpt, Dean Valentine recalls 2010s left-wing criticisms of effective altruism, including the charge that charity masks deeper problems such as capitalism, and appears to link them to arguments made about AI safety work. Only the opening was available; the main argument is not known.