I worked with autonomous vehicles a bit, and one of the elements of machine learning that jumps out from that time is the large investment in structuring the neural net and deciding overall architecture, as well as curating a good data set for training.
Notably, the thing that can be made to be quite fast is the training itself. This may not be that surprising, but in the context of the human mind, it's kind of amazing...it very much reminds me of the scenes in the first Matrix film where the characters can learn a skill (like kung fu or flying a helicopter) in a matter of seconds. We're not quite to that "instant expert" effect with AI, but we're not far off:
"It takes about an hour for the agent to learn to drive around a track. It takes about four hours to become about as good as the average human driver. And it takes 24 to 48 hours to be as good as the top 1% of the drivers who play the game."