A prize-winning training method promised a reward that couldn't be gamed -- then someone read the code and found it flipping a weighted coin whenever the real signal went missing. We trace that silent failure through three papers where the number on the label and the signal underneath quietly disagree: a self-play reward that grades itself, robots that act on the touch they're about to cause instead of the one they just felt, and a randomized study showing AI help can leave you worse at the very skill you're supposed to be learning. One useful takeaway for tomorrow: when you're learning, ask AI to explain, not to solve.

Chapters

0:00 The Reward That Made Up Its Own Answers
1:05 The Headlines -- Everybody Shipped, Nobody Shipped the Receipt
6:08 The Signal Underneath
7:05 The Reward Function That Graded Itself
11:25 Robots That Feel the Future, Not the Present
15:46 The Productivity Boost That Costs You the Skill
19:05 Read the Receipt, Not the Label

Links

The Reward That Made Up Its Own Answers -- https://arxiv.org/abs/2607.23802
Robots That Feel the Future, Not the Present -- https://arxiv.org/abs/2607.23782
The Productivity Boost That Costs You the Skill -- https://arxiv.org/abs/2601.20245