undefined | Better HN

0 pointsdgacmu2y ago0 comments

I didn't downvote you, but it seems like a pretty nitpicky question to ask at the "we demonstrated that the thing worked and didn't melt" stage.

It shouldn't have to compete - yet. Your question doesn't get at the more interesting one, which is, "will it eventually?"

0 comments

2 comments · 1 top-level

InDubioProRubio1y ago· 1 in thread

On contrair el capitan: It is trained on human designs and design processes. It has knowledge on physics and flow mechanics + propulsion simulations baked into it. It should outcompete a human, who gains that intuition by try and terror.

dgacmuOP1y ago

There's a saying from Rod Brooks that I quite like: "Simulations are doomed to succeed."

The hard part is making sure the simulation captures enough of the relevant factors and that the physics/etc simulation is accurate enough.

j / k navigate · click thread line to collapse