Aleten news and researchhttps://aleten.com/news/Sourced AI research and Ackren updates.en-gbEvidence, explanations and what changed at Ackrenhttps://aleten.com/newsletters/september-2026-evidence-and-explanationshttps://aleten.com/newsletters/september-2026-evidence-and-explanationsA September research digest: an inspectable teaching example, the limits of replay, and recent work on testing AI explanations.Fri, 25 Sep 2026 00:00:00 GMTTeaching a conversational engine, with the evidence visiblehttps://aleten.com/news/teaching-with-visible-evidencehttps://aleten.com/news/teaching-with-visible-evidenceA recorded local example shows Ackren deriving an answer from a statement and a rule. The prototype remains bounded and has not completed release qualification.Fri, 25 Sep 2026 00:00:00 GMTRepeatable is not the same as correcthttps://aleten.com/news/repeatable-is-not-correcthttps://aleten.com/news/repeatable-is-not-correctAckren’s recorded evaluation separates repeatable execution from correct interpretation. That distinction matters when assessing an AI explanation.Fri, 25 Sep 2026 00:00:00 GMTTo test an AI explanation, change the evidence it citeshttps://aleten.com/news/test-the-evidence-in-ai-explanationshttps://aleten.com/news/test-the-evidence-in-ai-explanationsA September preprint tests visual explanations through controlled image changes. The wider lesson is to test explanations against evidence, with clear limits on what a passing test proves.Fri, 25 Sep 2026 00:00:00 GMTVerification starts before the solverhttps://aleten.com/news/verification-starts-before-the-solverhttps://aleten.com/news/verification-starts-before-the-solverResearch on language-to-planning systems highlights a difficult boundary: a formally checked result depends on what the system translated from the user’s request.Fri, 25 Sep 2026 00:00:00 GMT