You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
Given infinite data, you can solve arbitrarily complex problems with a "dumb" model (such as a densely-connected neural network). However, when it comes to difficult real-world problems like self-driving, "lots of data" is very different from "infinite data". The long tail is fathttps://twitter.com/fchollet/status/1032444572306042880 …
There is no realistic amount of data that covers "everything you can encounter" in the real world. That's what intelligence is for: to be able to handle the real world despite having only seen a very small sample of what may happen. Intelligence is extreme generalization.
Also, ignoring/disregarding things. Like my painter ancestors said, "What I left out matters too."
Is that a rebuttal to Tesla's claim that they're on the verge of true self-driving ?
Do you think that driving is a problem that can be generally solved with machine learning and finite data?
Thx very insightful
I would guess Musk is prepared to sacrifice some lives to obtain the missing training data. It’s still probably lower risk than driving yourself.
If an AI car kills the wrong person then the executive of the company responsible may learn the hard way that hitmen are cheaper than lawyers. This self interest issue should guide them to be as prudent as possible, however if any are psychopaths they will be reckless anyway.
Intelligence must surely also have a preemptive simulation component? Running multiple scenarios with resources allocated based on the probability of a scenario following current sensor state. This generates patterns with action recommendations attached & best match is acted on.
Using an understanding of the situation, including its constituent elements, it can explore / simulate / predict / "prelive" each possible outcome, and then take action based on its significance (meaning) e.g. to try to avert an outcome that's deemed bad / undesirable.
You can represent meaning for machines using contemporary linguistics #RRG and #patomtheory . Meaning provides common sense and the appearance of logic - intelligence.
The example i give to people about self driving cars is if a piano falls from a building (like they do in the movies). It’s highly improbable, yet it could happen, and a human brain could consider how to operate (brake or accelerate). But a data fed AI? I don’t think so.
AI/ANN using radar and cameras at 2000fps would have better chance of detecting a large rapidly moving object moving into and blocking driveable space and output braking force faster than the human can even blink. Bad example.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
