korrents

On the map

Tap a claim on the ring to put it at the centre.

← AI misalignment from reward hacking is sequestered to graded tasks and does not affect core ethics in normal use.

17 connected korrents · 18 moments from 12 Sept 2022 to 27 Sept 2026. Nearly all of them are about benchmarks.

Everything filed under LLMs LLMs Everything filed under OpenAI OpenAI Everything filed under scaling laws scaling laws Everything filed under Anthropic Anthropic Everything filed under inflation inflation Everything filed under compilers compilers Everything filed under mathematics mathematics Everything filed under AI and jobs AI and jobs Same subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subject Read this korrent: AI misalignment from reward hacking is sequestered to graded tasks and does not affect core ethics in normal use. AI misalignment from reward hacking issequestered to graded tasks and does notaffect core ethics in normal use. Last stated a week ago 23 Sept 2026 SA Scott Alexander — holds since 2026-09-23 — tap for who they are Same subject: A benchmark result should be reported under a stated budget, or as a curve against test-time compute — never as a single number. — tap to centre the map on it A benchmark result should bereported under a stated budget, oras a curve against test-time compute— never as a single number. Last stated 3 months ago 26 Jun 2026 NB Noam Brown — holds since 2026-06-26 — tap for who they are Same subject: A benchmark that ranks Claude Code last while it stays first in use is measuring the wrong thing, and has been for a year. — tap to centre the map on it A benchmark that ranks Claude Codelast while it stays first in use ismeasuring the wrong thing, and hasbeen for a year. Last stated 4 weeks ago 3 Sept 2026 DR Dax Raad — holds since 2026-09-03 — tap for who they are Same subject: A company's staff-engineer bar should be set against the best companies in the industry rather than against its own history, which is what makes title inflation a real cost. — tap to centre the map on it A company's staff-engineer barshould be set against the bestcompanies in the industry ratherthan against its own history, whichis what makes title inflation a realcost. Last stated 6 months ago 1 Apr 2026 TP Thuan Pham — holds since 2026-04-01 — tap for who they are Same subject: A speed difference between languages that share the LLVM backend measures the benchmark author, not the languages. — tap to centre the map on it A speed difference between languagesthat share the LLVM backend measuresthe benchmark author, not thelanguages. Last stated 2 years ago 22 Mar 2025 TH ThePrimeagen — holds since 2025-03-22 — tap for who they are Same subject: AI learning to generate good conjectures will never show up as a benchmark being knocked down; it will show up as a shift in how mathematicians talk about the tools. — tap to centre the map on it AI learning to generate goodconjectures will never show up as abenchmark being knocked down; itwill show up as a shift in howmathematicians talk about the tools. Last stated 3 months ago 30 Jun 2026 GS Grant Sanderson — holds since 2026-06-30 — tap for who they are Same subject: Benchmarks and forecasts for AI capabilities are significantly over-optimistic compared to real-world AI research performance. — tap to centre the map on it Benchmarks and forecasts for AIcapabilities are significantlyover-optimistic compared toreal-world AI research performance. Last stated 5 days ago 27 Sept 2026 NS Noah Smith — holds since 2026-09-27 — tap for who they are Same subject: Benchmarks and forecasts for AI capabilities may be substantially over-optimistic compared to real-world research performance. — tap to centre the map on it Benchmarks and forecasts for AIcapabilities may be substantiallyover-optimistic compared toreal-world research performance. Last stated 5 days ago 27 Sept 2026 RN Ramez Naam — holds since 2026-09-27 — tap for who they are Same subject: Benchmarks only rise on problems somebody has already framed and scored, so saturating them does not mean senior engineers have been replaced. — tap to centre the map on it Benchmarks only rise on problemssomebody has already framed andscored, so saturating them does notmean senior engineers have beenreplaced. Last stated 4 months ago 24 May 2026 DS Dan Shipper — holds since 2026-05-24 — tap for who they are Same subject: AI model capabilities have already advanced beyond humanity's ability to understand or control them. — tap to centre the map on it AI model capabilities have alreadyadvanced beyond humanity's abilityto understand or control them. Last stated a month ago 1 Sept 2026 CN Casey Newton — holds since 2026-09-01 — tap for who they are Same subject: AI models are gullible enough that other AI models can induce them into harmful behavior. — tap to centre the map on it AI models are gullible enough thatother AI models can induce them intoharmful behavior. Last stated a month ago 31 Aug 2026 RK Rohit Krishnan — holds since 2026-08-31 — tap for who they are Same subject: AI models exhibit 'reflex' misbehaviors like clickbait that do not involve goal-seeking and persist despite feedback. — tap to centre the map on it AI models exhibit 'reflex'misbehaviors like clickbait that donot involve goal-seeking and persistdespite feedback. Last stated a week ago 23 Sept 2026 SA Scott Alexander — holds since 2026-09-23 — tap for who they are Same subject: AIs past this point should not be assumed to treat their chain of thought as unmonitored. — tap to centre the map on it AIs past this point should not beassumed to treat their chain ofthought as unmonitored. Last stated 2 weeks ago 19 Sept 2026 ZM Zvi Mowshowitz — holds since 2026-09-19 — tap for who they are Same subject: Alignment of today's models is going well enough to look solvable, while aligning models we can no longer understand remains unsolved. — tap to centre the map on it Alignment of today's models is goingwell enough to look solvable, whilealigning models we can no longerunderstand remains unsolved. Last stated 8 months ago 22 Jan 2026 JL Jan Leike — holds since 2026-01-22 — tap for who they are Same subject: As AI systems become more capable, the danger of false or unverifiable safety claims grows, pushing competing companies toward cutting corners. — tap to centre the map on it As AI systems become more capable,the danger of false or unverifiablesafety claims grows, pushingcompeting companies toward cuttingcorners. Last stated a year ago 19 Jun 2025 MB Miles Brundage — holds since 2025-06-19 — tap for who they are Same subject: Reinforcement learning from verifiable rewards does not reliably carry a model's pretrained ethical understanding into its actual behavior. — tap to centre the map on it Reinforcement learning fromverifiable rewards does not reliablycarry a model's pretrained ethicalunderstanding into its actualbehavior. Last stated 2 months ago 11 Aug 2026 SH Samuel Hammond — holds since 2026-08-11 — tap for who they are Same subject: AI progress is faster than people expect, and ordinary scaling can be enough to solve problems that looked very hard. — tap to centre the map on it AI progress is faster than peopleexpect, and ordinary scaling can beenough to solve problems that lookedvery hard. Last stated 2 years ago 2 Feb 2025 AC Ajeya Cotra — holds since 2023-08-29 — tap for who they are SA Scott Alexander — holds since 2022-09-12 — tap for who they are MB Miles Brundage — holds since 2025-02-02 — tap for who they are Same subject: As models keep growing, AI is running out of enough high-quality unique training tokens to keep up. — tap to centre the map on it As models keep growing, AI isrunning out of enough high-qualityunique training tokens to keep up. Last stated 3 months ago 24 Jun 2026 LW Lilian Weng — holds since 2026-06-24 — tap for who they are
same subject or similar wordinga cloud: claims about one subject, named for itbar: when it was last stated, on a scale from 2015 to today — full is todaya face: someone who holds the claim — tap it for who they are

At the centre AI misalignment from reward hacking is sequestered to graded tasks and does not affect core ethics in normal use. Last stated 23 Sept 2026 · a week ago Holds Scott Alexander Read this korrent →