We don't have 100% autonomous trains yet, yet we're supposed to have self driving cars next year?!

This is terribly naive, and perhaps stupid. Namely equating a self checkout machine to a driving car...Self driving cars in most environments are safer than human drivers RIGHT NOW. Your last two sentences apply to humans as well. Humans can perform far less tasks than computers can before becoming compromised. Computers struggle with the types of visual pattern recognition that humans are remarkably good at. This isn't some magical trait, computers will be better than us at it sometime soon. Neither a robot nor a human will ever be 100% in all situations, but the robot will be perfect in 99.9999% wheres the human only in 99%Dealing with Bad Wigins takes years off my life...

Bad Wigins wrote:

It's more easily explained as complexity compromises performance. The more decisions an AI makes, the more things can and will go wrong.

The only way a robot can be 100% is if you nail it down and provide it an environment where nothing unexpected ever happens. Checkout bots shut down until that unexpected item is out of the bagging area.

Argumentum ad hominem and straw man will definitely age you if you use them against me. Not only do you fail to rebut my statements, but you invite me to retaliate in kind, and I'm better at it.

The checkout robot isn't a comparison to autos, it's a contrast. The point went well above your head, for some reason, that even a simple, fixed robot like a checkout machine is thrown off by the unexpected. A complex system much more so.

Now to go ad hominem, it's obvious you have never programmed AI, or probably anything, and are under the illusion that you can just write code to deal with every contingency. What you don't understand is that the code itself creates uncertainty. The more code you add, the greater the chance that it will do something you did not intend. That is not safe and the public will never accept it on the road.

It's clear you want the luxury of a self-driving car so badly you are willfully blind to the danger they pose to human life. That is a typical bourgeois pig attitude. Hands on the wheel or hand in your license.

DiscoGary wrote:No. I am saying that if 10,000 autonomous cars were crashed by hackers in one year, then EVERYONE would be demanding that they be outlawed except for you and Xanax. It takes mental gymnastics to imagine how the tech could survive such a hit.

I'm also saying that the designers have been very lax on this so far, and if that doesn't change they will have big problems selling this.

DiscoGary and the New York Times agree. Heads explode.

wejo wrote:So the question becomes when will we have fully 100% self driving cars transporting us around?

According to this, 2025 or after.

Of course these cars have been on the road already for years, so we know they can do it, and that they do it currently with few accidents, according to the data they release. I don't know that they have been honest about that, because you cannot trust corporations. However, the fact is that aside from proper maintenance, which many of us do not do, the major cause of accidents is not merely human error but human aggressiveness. In other words, the vast majority of non-equipment related accidents have to do with tolerances being cut way down by aggressive driving. For instance, if you turn left into the next lane and hit a car in that lane, probably you would have seen it coming if it wasn't going 80 in a 55. Think of how many fender benders are caused by failure to observe proper following distances. A very low percentage of people observe them here. Each of you that can say they have never got into an accident can testify to the way that proper driving behaviors would cut down by 99% on accidents. Of course, bad driving on the part of others when you are not driving badly can force accidents, but that's the point. People could slash accidents by driving conscientiously and wisely.

xcvxcvxc wrote:

Of course these cars have been on the road already for years

they have not, they are supervised by humans, and not ones distracted by other activities which is highly illegal.

don't know that they have been honest about that, because you cannot trust corporations.

Unless you choose to be naive, you can presume that they are not honest.

However, the fact is that aside from proper maintenance, which many of us do not do, the major cause of accidents is not merely human error but human aggressiveness. In other words, the vast majority of non-equipment related accidents have to do with tolerances being cut way down by aggressive driving.

Red herring. This doesn't show that humans are unsafe drivers, on the contrary it shows that things have to go really wrong for humans to crash.

I rarely resort to this, but at this point I'm going to say that anyone who has never programmed an AI simply has no idea what they're talking about. It only works in a sandbox, not running free in the real world.

wejo wrote:So the question becomes when will we have fully 100% self driving cars transporting us around?

When car manufacturers can solve the Riddle of the Kangaroo.

Will never work in Australia.

Kangaroos are like no other large animal in that they don’t take steps, but bound along from spot to spot, “flying” for just a second at a time, landing, and then taking off again.

Level 5 autonomous vehicle technology is nowhere close to being ready for prime time-

https://jalopnik.com/hackers-could-confuse-driverless-cars-by-manipulating-w-1797017370

Humans, i.e. most people in this thread, can really only think in linear terms.

AI and computer vision is moving much, much faster than linear.

40 years to L5 autonomous cars is complete BS.

Anyone alive more than 15 years should be able to look at the history of young, essential, breakthrough tech and see this for themselves.

But, tech writers need to write though-pieces, so here we are debating.

The House unanimously approves self-driving car law:

https://www.reuters.com/article/us-autos-selfdriving/house-unanimously-approves-sweeping-self-driving-car-measure-idUSKCN1BH2B2

The U.S. House on Wednesday unanimously approved a sweeping proposal to speed the deployment of self-driving cars without human controls and bar states from blocking autonomous vehicles.

The bill now goes to the Senate and would allow automakers to obtain exemptions to deploy up to 25,000 vehicles without meeting existing auto safety standards in the first year, a cap that would rise to 100,000 vehicles annually over three years.

Apparently there's not one single representative that has any idea how dangerous this is.

DiscoGary wrote:

The House unanimously approves self-driving car law.

A unanimous vote will provide lots of political cover when the first people are killed by autonomous cars. Personally, I'd rather not let strangers experiment with my life and safety when the technology is still at such an early stage.

The killer cars are coming:

Driverless technologies are not to be coddled or promoted. Their development should be slow and sober. If the evidence warrants, their development should be stopped altogether. We are weaponizing our cars and trucks for use against us. It’s madness.

A better story:

https://www.nytimes.com/2018/03/19/technology/uber-driverless-fatality.html

Let's wait until all the facts come out in that story before we begin crucifying autonomous vehicles.

The Russians will start hacking these cars and causing them to drive on the wrong side of the road.

Harambe wrote:

Humans, i.e. most people in this thread, can really only think in linear terms.

AI and computer vision is moving much, much faster than linear.

40 years to L5 autonomous cars is complete BS.

Anyone alive more than 15 years should be able to look at the history of young, essential, breakthrough tech and see this for themselves.

But, tech writers need to write though-pieces, so here we are debating.

Anyone alive more than 15 years knows that self-driving cars have been 'right around the corner' for many years now...

Silicon Valley prevailing wisdom seems to be that any problem can be solved if you set it up as a machine learning problem and then throw enough computing power at it. I have yet to see any AI breakthrough (in the unpredictable real world, not on a chess board) that I am truly impressed with.

I have gotten more pessimistic about totally autonomous cars by the year. Google was said to have them working 5+ years ago but they still don't seem to work in the wide variety of conditions in the real world.

Door knocker wrote:

Let's wait until all the facts come out in that story before we begin crucifying autonomous vehicles.

But who will do the investigation? Who can we believe?