korrents

On the map

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

← Interpretability agents are useful for evaluating techniques because…

17 connected korrents · 15 moments from 13 Feb 2023 to 12 Nov 2026.

Everything filed under AI agents AI agents Everything filed under metascience metascience Everything filed under coding agents coding agents Everything filed under AI alignment AI alignment Everything filed under LLMs LLMs Everything filed under neural networks neural networks Same subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subject Read this korrent: Interpretability agents are useful for evaluating techniques because they are reproducible in a way human researchers are not. Interpretability agents are useful forevaluating techniques because they arereproducible in a way human researchersare not. Last stated 11 months ago 20 Dec 2025 NN Neel Nanda — holds since 2025-12-20 — tap for who they are Same subject: No single interpretability technique is reliable enough alone to justify a verdict like a model being unsafe; decorrelated techniques are needed. — tap to centre the map on it No single interpretability techniqueis reliable enough alone to justifya verdict like a model being unsafe;decorrelated techniques are needed. Last stated 11 months ago 20 Dec 2025 NN Neel Nanda — holds since 2025-12-20 — tap for who they are Same subject: Complete reverse-engineering of neural networks is a less promising research direction than studying model biology and applying interpretability usefully — tap to centre the map on it Complete reverse-engineering ofneural networks is a less promisingresearch direction than studyingmodel biology and applyinginterpretability usefully Last stated a year ago 19 Aug 2025 NN Neel Nanda — holds since 2025-08-19 — tap for who they are Same subject: AI agents cannot be perfectly isolated if they are to be useful, because their usefulness depends on access to information. — tap to centre the map on it AI agents cannot be perfectlyisolated if they are to be useful,because their usefulness depends onaccess to information. Last stated 2 months ago 30 Sept 2026 MG Matthew Green — holds since 2026-09-30 — tap for who they are Same subject: Using AI agents to do a task does not by itself increase a person's own pace of understanding, intuition, or taste for that work. — tap to centre the map on it Using AI agents to do a task doesnot by itself increase a person'sown pace of understanding,intuition, or taste for that work. Last stated 3 months ago 12 Aug 2026 NL Nathan Lambert — holds since 2026-08-12 — tap for who they are Same subject: The best interpretability deep dives combine several kinds of evidence, since no single technique is trustworthy enough on its own. — tap to centre the map on it The best interpretability deep divescombine several kinds of evidence,since no single technique istrustworthy enough on its own. Last stated 11 months ago 27 Dec 2025 NN Neel Nanda — holds since 2025-12-27 — tap for who they are Same subject: AI is good at giving artifacts that are not necessarily grounded in understanding of the system, which works for prototypes and one-off experiments but leaves you lost the moment you need to debug. — tap to centre the map on it AI is good at giving artifacts thatare not necessarily grounded inunderstanding of the system, whichworks for prototypes and one-offexperiments but leaves you lost themoment you need to debug. Last stated 8 months ago 25 Mar 2026 FG Flora Guo — holds since 2026-03-25 — tap for who they are Same subject: Anthropic now uses mechanistic interpretability techniques to audit new models for deception and scheming before releasing them. — tap to centre the map on it Anthropic now uses mechanisticinterpretability techniques to auditnew models for deception andscheming before releasing them. Last stated 11 months ago 1 Jan 2026 DA Dario Amodei — holds since 2026-01-01 — tap for who they are Same subject: Current AI models do not actually understand things and still hallucinate. — tap to centre the map on it Current AI models do not actuallyunderstand things and stillhallucinate. Last stated a year ago 13 Oct 2025 SG Saul Griffith — holds since 2025-10-13 — tap for who they are Same subject: “Trust science” is a confused instruction: what deserves trust is the process, not any particular thing it has concluded so far. — tap to centre the map on it “Trust science” is a confusedinstruction: what deserves trust isthe process, not any particularthing it has concluded so far. Last stated 3 years ago 16 Aug 2023 JC Jason Crawford — holds since 2023-08-16 — tap for who they are Same subject: A Cochrane Systematic review of 11 trials found no evidence of benefit and a small signal of harm for mechanical CPR devices. — tap to centre the map on it A Cochrane Systematic review of 11trials found no evidence of benefitand a small signal of harm formechanical CPR devices. Last stated 3 months ago 17 Aug 2026 JM John Mandrola — holds since 2026-08-17 — tap for who they are Same subject: A follow-up clinical trial was unable to reproduce the earlier improvement in insulin sensitivity from blueberries. — tap to centre the map on it A follow-up clinical trial wasunable to reproduce the earlierimprovement in insulin sensitivityfrom blueberries. Last stated 2 months ago 6 Oct 2026 RP Rhonda Patrick — holds since 2026-10-06 — tap for who they are Same subject: A review process where authors decide whether and how to address comments without an editor is an unscientific sham. — tap to centre the map on it A review process where authorsdecide whether and how to addresscomments without an editor is anunscientific sham. Last stated a week ago 12 Nov 2026 AD Andrew Dessler — holds since 2026-11-12 — tap for who they are Same subject: Checking a paper's data and analysis is almost never done, though it may be the most important thing review could do. — tap to centre the map on it Checking a paper's data and analysisis almost never done, though it maybe the most important thing reviewcould do. Last stated 4 years ago 13 Feb 2023 AM Adam Mastroianni — holds since 2023-02-13 — tap for who they are Same subject: Economics journals have made large datasets and heavy statistics the price of making any causal claim at all, which has squeezed qualitative economic history out of the field. — tap to centre the map on it Economics journals have made largedatasets and heavy statistics theprice of making any causal claim atall, which has squeezed qualitativeeconomic history out of the field. Last stated a year ago 13 Oct 2025 AH Anton Howes — holds since 2025-10-13 — tap for who they are Same subject: "I don't know" becomes a kind of superpower when you can work with Claude Code to figure things out. — tap to centre the map on it "I don't know" becomes a kind ofsuperpower when you can work withClaude Code to figure things out. Last stated 8 months ago 19 Mar 2026 KP Kris Puckett — holds since 2026-03-19 — tap for who they are Same subject: A code-review agent works best with a clean context, sharing nothing with the coding agent beforehand. — tap to centre the map on it A code-review agent works best witha clean context, sharing nothingwith the coding agent beforehand. Last stated 7 months ago 22 Apr 2026 WY Walden Yan — holds since 2026-04-22 — tap for who they are Same subject: A coding agent does not learn from its mistakes the way a person does -- it repeats the same error indefinitely unless a human notices and writes it down. — tap to centre the map on it A coding agent does not learn fromits mistakes the way a person does-- it repeats the same errorindefinitely unless a human noticesand writes it down. Last stated 8 months ago 25 Mar 2026 MZ Mario Zechner — holds since 2026-03-25 — 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 Interpretability agents are useful for evaluating techniques because they are reproducible in a way human researchers are not. Last stated 20 Dec 2025 · 11 months ago Holds Neel Nanda Read this korrent →