The honest state of AI understanding — what's proven, what's marketing, and what's genuinely unknown.
What AI can actually do today (and what's marketing)
How large language models work (no PhD required)
The scaling hypothesis — will bigger always mean better?
What alignment research has and hasn't solved
The measurement problem — why benchmarks lie
Genuine unknowns that nobody is pretending to have answers to
Clear explanations of what AI can and can't do, written for smart generalists.
Read The History of AI →