His AI strategy that doesn't use IF statements sounds influenced from the Sussman & Radul paper the Art of the Propagator. In this related course you also learn how to program AI decisions based on pattern matching like him giving space to a cyclist and the AI later doing the same https://groups.csail.mit.edu/mac/users/gjs/6.945/
The first person to hack the iPhone is building a self-driving car
221–230 of 462 posts
Re: The first person to hack the iPhone is building a self-driving car
#222Earlier quoted context omitted.
Look at any automated warehouse these days. Pickers, stockers, warehouse managers..... Those are going by the way side.
Would you consider those machines AI?
strong AI? no- but then, that hasn't been achieved yet.
but weak AI? absolutely. Look at amazon's picker bots.
Re: The first person to hack the iPhone is building a self-driving car
#223If it passes the written and driving tests at the local DMV, should the car be given a driver's license?
no, because that's silly.
Re: The first person to hack the iPhone is building a self-driving car
#224So its more a fancy autopilot than a self driving car.
Its right there in the article, he is trying to beat Mobileye - a lane following technology Tesla rebranded as autopilot.
Re: The first person to hack the iPhone is building a self-driving car
#225So its more a fancy autopilot than a self driving car.
at this point, yes. clearly this is not the endpoint.
Re: The first person to hack the iPhone is building a self-driving car
#226Earlier quoted context omitted.
I cringed. Saying "the math is simple" -- ouch. The writer must have been barely able to suppress his glee when that one popped out.
I don't doubt it at all. Keep in mind that "simple" is relative, we have to ask "simple compared to what?" For lots of people, I be that the math involved in these neural nets is the most complicated math they've ever done. They would never say it is simple, because they themselves barely grasp it. But in my experience, topics in mathematics have a funny way of becoming very simple the moment you "graduate" to thinki…
While a good point that simple is relative, if you specifically look at deep neural networks we don't understand how training a non convex function converges with gradient descent - the fundamental component to create a usable model. In practice, it often works, and there are a few intuitions why this works. But its naive at best to claim that this is simple. If it was we would understand it better :)
Re: The first person to hack the iPhone is building a self-driving car
#227Earlier quoted context omitted.
if we don't have laws currently, we sure as hell need them. i can't imagine letting everybody try their own self-driving software.
Millions of drunk people drive on the roads. People can buy assault weapons, with no training and background checks. People who make self-driving software should be the least of our concerns.
There are laws against drunk driving (and harsh penalties for those that are caught), and you can't buy a firearm without a background check from a dealer (with more states requiring gun show dealers perform background checks now, too).
People building untested self driving cars is an entirely legitimate concern.
Re: The first person to hack the iPhone is building a self-driving car
#228Earlier quoted context omitted.
what's with making text so thin you can't even read?
I hope the design was originally with a different image. And then the image was changed, and it wasn't redesigned.
The text that isn't overlaying images is terrible too. It's too thin for subpixel rendering to look decent. There's not enough contrast for viewing on a TN LCD panel unless it's in the middle of the screen.
Re: The first person to hack the iPhone is building a self-driving car
#229Earlier quoted context omitted.
He has a company: http://comma.ai .
and http://reactions.co
I mean http://comma.ai is the self-driving car company. (Reactions was a hackathon project, not a real co.)
Re: The first person to hack the iPhone is building a self-driving car
#230Earlier quoted context omitted.
Completely agree. There's absolutely no way you could use this approach for level 4 autonomy. Level 3, fine, but not level 4.
In the video, he says his company is only targeting Level 3.
Ah I didn't see that. Fair enough. I still agree with the parent comment that the (important) edge cases like avoiding a crash will not be well handled potentially.