The CEO has no Robotaxis: my retort to Adam Jonas and Morgan Stanley. Please share.
There was an unfortunate but minor typo in the paper where I misspelled Yoshua Bengio's last name. I've since corrected it and attached an updated page 4 to this reply.
@seth1620 wen you get time. check Dr Shasua's tweet and concomitant blog (linked). you'll like the similar points. he's real engineer who can lecture for hours deep into the tech versus ewon's 'talking points' *btw very unusual for him to approach TSLA head-on. x.com/AmnonShashua/s…
@seth1620 wen you get time. check Dr Shasua's tweet and concomitant blog (linked). you'll like the similar points. he's real engineer who can lecture for hours deep into the tech versus ewon's 'talking points' *btw very unusual for him to approach TSLA head-on. x.com/AmnonShashua/s…
@seth1620 I agree with this but would go further in ridiculing claims current Teslas can ever be robotaxis. If Tesla ever certifies FSD for robotaxi use, they take on the liability for potentially millions of cars! The crashes and law suits begin immediately and don’t stop til bankruptcy.
@seth1620 Beautiful formulation: “In truth there are an infinite number of permutations for the movement of objects through the world and thus an infinite number of edge cases.” My is: closed systems don’t work in an open discrete and indeterministic universe …
@seth1620 Hmm, not sure I necessarily agree with the statement that autonomy is arbitrarily too Kolmogorov complex to train in short enough timeframes. While it is hard to meaningfully anticipate process success rates vs. model/dataset size, I don’t think waiving around infinity is useful.