2 ms·
> The set of people working on automobile R&D who are good candidates to work on "generative AI projects" is so small that it's hardly worth mentioning. Maybe
by doctorpangloss 3y ago
> The set of people working on automobile R&D who are good candidates to work on "generative AI projects" is so small that it's hardly worth mentioning.
Maybe the entire, most valuable part of the team is working on synthetic data, since at least 2020, and maybe as early as 2017.
Apple’s Hypersim dataset is a good example of how essential it is.
Synthetic data delivered Sora. It delivered Parallel Domain, an autonomous driving data startup. It’s been the state of the art idea for longer than GPT 3 has been around.