The BaseVideoMachine Learning Street Talk1 min3,262 views
LLMs Cosplay Understanding (Jeremy Howard)
According to the description, in a one-minute short from a longer interview, Jeremy Howard of fast.ai and Answer.AI says both sides of the debate on whether LLMs understand are right: they imitate understanding, which suffices for many tasks until the pretence fails.
Why it matters
The same thing, for different readers.
What it means for you
Jeremy Howard says AI chatbots can seem to understand without truly doing so, which is fine for many tasks but may fail at some point.
For researchers
Howard's framing, drawn from Searle's Chinese room, suggests asking where behaviour that imitates understanding stops being a good enough substitute for it.
The question it raises
The clip revives the Chinese room question: does acting as if you understand amount to understanding?
The record
- Questions
- What values should AI have, and could it have a mind of its own? · How fast is AI advancing, and where is it heading?
- People
- Jeremy Howard (Speaks, Co-founder of fast.ai and Answer.AI)
- What we read
- We have read only the description the publisher gives. No quotations until someone reads the source itself.
- Found by
- The Radar, run of November 12, 2026, via
yt-machine-learning-street-talk-youtube(tier 1: the source’s own channel or an official record) - Cite this
- Machine Learning Street Talk, “LLMs Cosplay Understanding (Jeremy Howard)”, November 11, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/mlst-2026-11-11-howard-llms-cosplay-understanding