I have seen Fridman's pitch. There is a difference between understanding current methods and how succesful they are in various domians. In this one GL and DL are nowhere near where they need to be. Setting that aside is the dev approach. And that is my biggest issue. Not possible to rely on the real-world from a safety, time or money POV.

Musk Funded OpenAI Quits Working on Robots (Autonomous Vehicles)

https://imispgh.medium.com/musk-funded-openai-quits-working-on-robots-autonomous-vehicles-81a0abb5e479

The future of deep learning, according to its pioneers - https://bdtechtalks.com/2021/07/01/deep-learning-future-bengio-hinton-lecun/

The Autonomous Vehicle Industry can be Saved by doing the Opposite of what is being done now

https://medium.com/@imispgh/the-autonomous-vehicle-industry-can-be-saved-by-doing-the-opposite-of-what-is-being-done-now-b4e5c6ae9237

And Tesla's sensor system is fatally flawed.