“A violation of Richard Sutton's "bitter lesson" — the idea that more compute and data will always outperform human algorithmic ingenuity — is the biggest risk to the AI investment thesis.”