Don’t let the exponential gains in ability fool you: there are fewer real grand plans in AI (or frankly any field of human endeavor) than you think. Companies are pivoting as the market changes, sometimes models are really good or really bad, and a lot of stuff is just improvised