SoylentNews is people
SoylentNews
from the but-will-it-stop-for-coffee++? dept.
Arthur T Knackerbracket has found the following story:
Udacity President Sebastian Thrun speaking at Vanity Fair's New Establishment Summit.
Prepare for your car to become an intellectual giant -- and for you to like it.
In a highly optimistic forecast at the Vanity Fair New Establishment Summit in San Francisco, computer scientist Sebastian Thrun said artificial intelligence will radically reshape our lives for the better.
"In the last 200 or 300 years, we have made ourselves into superhumans," able to plow a field a thousand times faster than our ancestors, fly across the Atlantic Ocean and talk to a person in Australia, he said. Artificial intelligence will take us to the next step: "Rather than replacing our muscles, we're going to be making our brains stronger."
That'll start with artificially intelligent cars, said Thrun, who rose to Silicon Valley fame in his former job leading Google's self-driving car project.
"All the unborn cars get born with the full wisdom of their forefathers. AI cars will outpace all of us because they can learn faster," said Thrun, still a Stanford professor and now president of online learning site Udacity.
Artificial intelligence is spreading like wildfire across the technology industry, screening out junk email, labeling our photos, translating foreign languages and helping us type faster. But not everybody is so sanguine about the possibility of AI machines taking over high-skilled jobs.
(Score: 3, Interesting) by Unixnut on Sunday October 23 2016, @11:01PM
It's like asking a Catholic Priest whether they think Catholicism is awesome.
The guys career, years of work, current and future earnings are at stake. He isn't exactly going to say "It's a fad that will die out in 5 years" now will he?
Besides, everything I have seen so far, I would not call "Artificial intelligence". The intelligence in the machine is still imparted by humans. The humans to program it (even if it means setting up the neural net structure), humans to guide it, filter information and spoon feed the neural network, and tell it the results they want to achieve.
Yes, now we can let the machine decide how to get to the answer given all the above, but the intelligence is still all in the human. And despite all this work, the neural networks can give completely random output, which needs correcting by (you guessed it) humans. Especially for edge cases. Neural networks are more feedback based pattern matching systems than real Intelligence (aka "reinforcement learning systems").
We are a long way off from real AI, where a machine is able to learn by itself with nothing more than a command instruction. Don't get me wrong, the new class of thinking machines are very impressive, and I think they will have a modest impact in the world.
However the technology media seems to have gone into a rabid frenzy. If you ask them we are just a few years away from C3PO, robot butlers and some post-capitalist utopia where robots will do everything for us.
Sorry, but no, we are nowhere near that point (if it is even possible to reach it). Plus I suspect that as we get more powerful technology and AI-like machines, they are more likely to be used against the masses than for them. Think less "robot butler" for you, and more "robot policeman" to keep you in check and working hard for those in power.
(Score: 2, Interesting) by Anonymous Coward on Monday October 24 2016, @12:01AM
I once asked a priest why Catholicism was best. He replied, "It's not the best, otherwise they wouldn't be leaving in droves. They find the services boring and tedious; the teaching, complicated; the history, wanting; the hypocrisy, untenable; and the teachings, old-fashioned. No people always look for and find better elsewhere. But they forget, just because something feels good doesn't mean that it is the best. People have fled the church for comfort for centuries and will continue to do so because we don't offer what is "best" but offer the most reliable way to the "Truth."
The way he rattled that off with such little time to prep himself has me somewhat convinced that he had that filed in his brain with other common questions.
(Score: 0) by Anonymous Coward on Monday October 24 2016, @02:47AM
(Score: 2) by VLM on Monday October 24 2016, @11:37AM
otherwise they wouldn't be leaving in droves
A couple decades ago with my wife going thru the whole catholic marriage counseling process, we met an interesting priest who's point of view on people leaving was it didn't matter because theres more than a billion Catholics plus about a billion more "not Catholics but more or less Christians" and given two billion people if you can't find your place somewhere thats because you failed, not because the church failed. Don't like it here, find another chuch, there's only like 100 million others. Of course he put it in somewhat nicer language and I think included a bible quote for fun etc etc.
Anyway its relevant in that if you consider F'ing thermostats to be AI, ( thermostats! ) then its hard to say AI will ever die or become irrelevant. There's people out there in the business world BTW who are so ignorant they think PID control and newtons method are both AI. If you water down the definition of AI enough, then anything with a transistor in it is AI, which is a hell of a lot.
(Score: 2) by VLM on Monday October 24 2016, @11:54AM
AI ... will outpace all of us because they can learn faster
Why? Its a religious belief so it can't be questioned or reasoned with, but "why?" as usual kicks religious questions right in the balls.
If you're familiar with the jargon, for centuries everything from individuals to enormous voting collectives and everything in between have tried to implement Chinese Rooms (an AI term relating to philosophy of composition in programming) that implement business and government. They have been more effective than random dice rolling yet have often implemented amazing stupidity and evilness. Also they are legendarily slow, and the bigger and more "powerful" they are, the less agile and effective they always have been. Your five man town council can repave main street on time and under budget as a small weak Chinese Room AI where the participants act out their parts, but only the giant DoD can deploy a fighter plane that costs ten times too much and took three times longer than estimate to R+D.
Its an especially big problem for AI because no body is going to deploy a supercomputer cluster to project manage the repaving of main street in my little suburban city but naturally it'll be deployed against "cancer" or "poverty" or some damn thing where the largest and best managed Chinese Rooms in history are currently failing, so we'll give it a bigger Chinese Room than ever to totally F things up.
So if we can't even implement a human powered Chinese Room AI implementation of "dig uranium out of ground; make electricity" without repeatedly fucking up and working infinitely slowly and semi-incompetently, then why would a bunch on non-subject matter experts using generic instead of specialized tools be expected to do any better? On a small product run, who makes better horseshoes, a village blacksmith who's been horseshoeing since he was an apprentice or some jackass with a really fast computer and nothing else relevant to the problem?
I don't know how to emphasize this more strongly. We have "Chinese Room" style AI for centuries and it generally sucks ass and is called corporations and governments. We don't know how to organize a better "Chinese Room" so using fast transistors instead of neurons isn't likely to change anything, especially when the biggest historical F ups have never been caused by a processor speed limitation.
Its an evangelical christian idealization to imagine AI as heaven when its more likely to resemble GM in the 70s or the Reich in the early 40s or Genghis Khan's travel adventures. Just be straight with us that Jesus is coming and Thrun is his apostle and AI isn't going to bring heaven to earth it'll probably look a hell of a lot more like the Guinness corporation, Zimbabwe, or Exxon. With Guinness as the absolute best case irrational belief and a thousand year Reich more likely.
(Score: 1, Interesting) by Anonymous Coward on Monday October 24 2016, @01:10AM
> C3PO, robot butlers and some post-capitalist utopia
More likely Marvin and the marketing department of the Sirius Cybernetics Corporation. Need I remind you of, "A bunch of mindless jerks who were the first against the wall when the revolution came." (DNA)
I'm in with the idea of improving interfaces so that computers are better at what they do best. And I don't mean various versions of the MS Office "ribbon" that seem to be popping up everywhere, hiding all kinds of little-used features.
Will AI give us better office tools? If we get Clippy (or Eddie the shipboard computer) I may have to retire to the woods.