Teslaâs response to this brings up some very important points. Mainly:
getting a machine learning system to be 99% correct is relatively easy, but getting it to be 99.9999% correct, which is where it ultimately needs to be, is vastly more difficult.
I mean good on Geohotz for fiddling around with this stuff, but his system is nowhere near safe or accurate enough to be a commercial product (and likely never will be). The liability is astronomical. When companies full of Very Smart People with nearly-bottomless funding are still taking baby steps, itâs absurd to think that you can just jump in with a garage-based project and get the same results.
I predict much anguished hair-pulling over the notion that a free and independently motivated individual, rather than a wealthy, amoral corporation, has done something so potentially dangerous.
Also I predict calls for inane and counterproductive legislation.
HA HA HA! Best post of the day!
The gentleman is about 25, and may resent being called a kid.
Thatâs just a phase, heâll grow out of it.
I believe âwhippersnapperâ is the appropriate term.
So bizarre what is considered âjournalismâ at the moment. This is basically a PR release for a self-proclaimed ânext tech billionaireâ⌠The reporter doesnât seem to do even the cursory analysis of how this tech performs. We donât see the car do anything other than drive straight down the highway⌠Is it just a glorified cruise control? Does it handle city streets? Pedestrians? Neat DIY hack but far from the bluster the reporter lets his subject spew.
I get that maybe the guy wouldnât allow too much scrutiny but thereâs no reason Bloomberg should just be an unquestioning mouthpiece. Nothing I should be surprised of I suppose.
I think weâre having a palm-pilot moment here; this tech is further away than everyone thinksâŚ
Yes, but a plethora of garage-based projects indicates that the subject-field is ripe for innovation and tinkering.
It doesnât mean it scales, but sometimes the sideways , counter-intuitive cargo-cult approaches can pay off (if for no other reason than indicating an emerging, fanatical market, field-testing components).
Jesus, the level of arrogance he has is pretty nuts.
I suppose itâs possible that undertaking the task of automated driving requires some level of arrogance, but gosh itâs unpleasant to be around.
I mean: whatâs the likelihood that all the other people working on this have chosen the âwrong wayâ to solve the problem (and this young man has found the âright wayâ) VS this young man has found a way to solve some of the problems (but that wonât scale to the ever-so-important edge cases)?
Anyway.
Until I can sleep in the car and it can drive more safely than I can, I donât much care. So, that, what? Level 4 automation?
Wake me up when that happens.
In the field of cutting edge R&D, humility wonât get you far. You have to go for the âimpossibleâ. You may fail. You may fail again. You have to keep going. If it requires arrogance, so it be; we didnât conquer even the low Earth orbit with meekness.
What about catching a nap in the car while it inches its way through a clogged highway, or handling oneâs phonecalls (or game) in a drive-in queue? Thereâs a lot of not-so-critical tasks where even relatively simple automation can do a lot of work.
Thereâs a difference between confidence and persistence and arrogance. And a whole lot of room between that and meekness.
Napping in traffic? Nah.
I canât help but wonder what would happen if the resources currently directed at the automated driving âproblemâ were directed at the public transportation problem.
I have to rely on public transportation. It is fairly good here, to the point of being usable, but it still sucks, especially when it rains or snows or even just is generally cold. Or when hauling cargo. Or when tired.
Iâd prefer a vehicle to enter at the point of house door and get out at the destination, without the waiting and transfers and other logistical crap in between.
I hope he makes it on to the next season of âSilicon Valleyâ; huge potential for all sorts of antics.
I couldnât agree more! Itâs fucking crazy that everyone is trying to backwards engineer these automated systems to run on roads created ad hoc, 50+ years ago, rather than say, designing a decent automated public transit system that would work from the ground up. Itâs just a sad example of how throughly gutted the public infrastructure is that such an option isnât even on the table. Add to that all the vested interests throwing up roadblocks so that our bullet train may never come to pass.
Imagine automated transportation coupled with public transit. Suddenly you can expand how far public transit can go because you donât have drivers that get flustered memorizing routes or having to change over after x hours to keep from going into a kind of haze.
What about Uber(oid) with self-driving cars? Several Uber-like companies, mercilessly competing to keep the prices rock-bottom down?
Came here for this. If he were a she, I can only imagine the outrage if she were being called âgirlâ or âlittle missâ or some other dimunitiveâŚ
The problem is that public transport, when done well, scales far better than a CO2 blasting car for every person does. Plus, there are inbetween options, such as driving to a train station or using human powered vehicles like bikes for transfers etc.
Uses normal cameras, you say?
So, basically, itâll be useless when it rains.
