Oct 29, 2026 · 8 min · 6 segments
This fall, two randomized trials of AI tutors pointed in opposite directions. Maryland undergraduates who were offered a tutor in their course finished with lower grades, while secondary students in…
By default, it gave students direct explanations.
A second mode walked them toward an answer, and only two instructors ended the term using it.
A chatbot hallucinates when it states something false or unsupported with confidence.
A little over a year ago, researchers at OpenAI argued that the way models are trained and scored rewards the habit.
Most benchmarks count only right answers.
A model that guesses a stranger's birthday has a 1 in 365 chance of earning the point.
A model that admits it does not know earns nothing.
A trend report could then call it a trend, and a scenario built on that trend could reach a school district's budget discussion.
The tutor's default setting would have dropped out at the first step.
We call this a foresight hallucination.
A reading of the evidence sounds more certain at each step, while the evidence underneath it stays the same.
By default, it gave students direct explanations.
A second mode walked them toward an answer, and only two instructors ended the term using it.
A chatbot hallucinates when it states something false or unsupported with confidence.
A little over a year ago, researchers at OpenAI argued that the way models are trained and scored rewards the habit.
Most benchmarks count only right answers.
A model that guesses a stranger's birthday has a 1 in 365 chance of earning the point.
A model that admits it does not know earns nothing.
A trend report could then call it a trend, and a scenario built on that trend could reach a school district's budget discussion.
The tutor's default setting would have dropped out at the first step.
We call this a foresight hallucination.
A reading of the evidence sounds more certain at each step, while the evidence underneath it stays the same.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.