← Videos
Sounds Right Is It - AI Misstakes

AI from the Ground Up8 Nov 2026 · 4:23

Sounds Right Is It - AI Misstakes

An AI answer can sound helpful and still add a detail that the source never supplied. How do you catch that without rejecting everything it says?

From the episode description

An AI answer can sound helpful and still add a detail that the source never supplied. How do you catch that without rejecting everything it says? We examine a short workshop answer containing two claims: bring a laptop, and lunch is included. One is supported by the notes. The other is not. Then we introduce an outdated start time to show a different kind of mistake. In this episode: • What people mean by an AI hallucination. • Why confident wording is not an accuracy check. • How an accurate quotation can still come from the wrong source version. • What missing information does—and does not—allow you to conclude. • How to check a citation against the specific claim beside it. Workshop Notes and the displayed answers are fictional teaching examples, not a measured model error rate. The point is to practise an inspection method you can reuse. Take one answer you would normally accept quickly. Separate its factual claims, open the supporting passages, and keep any unresolved questions visible. Clear instructions help, but the evidence still needs checking. ResearchNode: AI from the ground up • Episode 04 of 10. Previous: How to Give AI Better Instructions Next: How AI Searches by Meaning: Embeddings Explained Presented by Michelle, an AI-generated presenter.

Descriptions and chapter markers are supplied by the channel. They are not a transcript or an independent check of every claim in the video.