<back
Everything I write
AI Safety in 2026 and Beyond: Non-Negotiables
Gametime
Your AIs don't do what you want. This is really bad
Secret Loyalties Likely Raise Remote-Influenceability
South Korea and the HBM Bottleneck
Publications
BAITBENCH: Measuring Agent Reward Hacking with Optional Shortcuts Planted in ML Tasks
Quantifying Phonosemantic Iconicity Distributionally in 6 Languages
Evaluating K-Fold Cross Validation for Transformer Based Symbolic Regression Models