Back to episode — Episode 2899 CWSA 07/16/25
Context —
no traffic on the roads. It would just be robots zipping around underneath the ground. Then you could order stuff that you wouldn't normally even bother ordering, like you could order a candy bar and maybe pay 25 or 50 cents to have it delivered. So anything you wanted would just sort of appear. Now at the moment, since they don't have pipes underneath everybody's house, the delivery goes to some…
← Previous segment →ry pipes. It's coming. So this is a real thing. It's already being built.
Elon Musk has confirmed that his xAI, which would be Grok 4, he says it's the smartest AI in the world and also it's going to be built into Optimus humanoid robots. But my question is this. How in the world is a large language model going to be safe to put in a robot? If the biggest problem with AI is hallucination, how does a robot learn not to do that? And does the hallucination apply to physical acts?
Now, I understand how AI can work in your fully self-driving car. If you have a bazillion hours of video of cars from the perspective of the car, then all you need is visual AI and apparently cars can drive themselves. But do you think you could have a human
Context —
oid robot that had seen enough video of the real world that it could navigate your house? It would just walk in your house and you'd say, "Robot, make me a sandwich," and the robot would know what your refrigerator looks like and where you keep the condiments and stuff. It would just figure it out. I don't know. I feel like that problem of hallucinating is unsolved and maybe unsolvable with any l…
Next segment → →