[jcannell]

This notion that AI/ML requires tons of data that only big tech companies have is at least partly a myth.

Big private datasets are important only in narrow domains - if you want to train an ANN to do ad prediction or something in medical imaging, sure.

But longer term and for the most valuable applications, the required data is plentiful and free. The data advanced robots (ex. self-driving cars) and AGI need to learn is all around us and it is free. Progress here is limited by compute/algorithms, not data.