Slashdot Mirror
Bill Gates: AI Is The 'Holy Grail' (mashable.com)
An anonymous reader writes: At the Code Conference on Wednesday, Bill Gates balanced his fears of artificial intelligence with praise. He talked about two of the challenges AI will pose: a loss of existing jobs, and making sure humans remain in control of super-intelligent machines. Gates, as well as many other experts in the field, predict there will be an excess of labor resources as robots and AI systems take over. He plans to talk with others about ideas to combat the threat of AI controlling humans, specifically noting work being done at Stanford. Even with such threats, Gates called AI the "holy grail" as he envisions a future "with machines that are capable and more capable than human intelligence." Gates said, "We've made more progress in the last five years than at any time in history. [...] The dream is finally arriving. This is what it was all leading up to."
Nobody will need more.
Because he provided us all with such reliable and well-engineered technology... er, gotten rich through monopolistic and otherwise ethically questionable business practices, and ruthlessly exploiting the network effect of a for him happy accident that left him with a huge installed base.
Anybody read his book? How many IQ points did you lose reading it?
Loss of jobs is the big one. An AI is not only not capable of killing humans, but would have nothing to gain from killing the people who maintain it. On the other hand, poor and unemployed people with nothing to lose will tear our society apart if that part grows large enough (as has been demonstrated numerous times throughout history) and I fear nobody seems to be taking this situation seriously. We need to find an alternative way to structure our society, and quickly, if we want AI that does all our work for us.
"Set a man a fire, he'll be warm for the rest of the night. Set a man afire, he'll be warm for the rest of his life."
Currency is an abstraction of labor, we use it to manage the effort put into things during trade - it's a lot more convenient than carrying around four cows and a goat. So, robots come along and take all the jobs? Well, no more scarcity of labor. And the systems of currency and capitalism we have grown so far get upended. They won't go out the window but they will see massive restructurings. If labor is not scarce, want a house? Go pick one down the street where the machines built fifty of them. Free. Because there was no scarce labor involved. Capitalism? Well, in a post scarcity economy the invisible hand that makes it go remains to be seen how that adapts. In the short term however, say ten to thirty years, a transition system where perhaps everyone gets a guaranteed minimum income until our society fully adapts to machines could help to minimize social upheaval over the machines taking all the jobs.
Shh.
"AI is the holy grail" - Bill Gates, 2016
"Two years from now, spam will be solved" - Bill Gates, 2004
"640K ought to be enough for anybody." - Bill Gates, 1981
Given the outcome of Gates' previous predictions, I think it's safe to presume that AI is and will never be the holy grail.
"Bill Gates"
"Expert in the field"
malicious snickering... :)
What's amazing is the number of people who, nowadays, and even in IT, don't understand that. AI has always been the Holy Grail. The difference is that, now, we are much closer to reach it than 20 years ago.
Slashdot, fix the reply notifications... You won't get away with it...
he started losing it
Our mission is to create him
The dangers of excessive individualism are nothing compared to the oppressiveness of excessive collectivism
Free, clean energy is. AI means the oligarchs get to remove more jobs from the masses, thus increasing suppression of dissent (until forced into revolution); but limitless energy means the world's population can all live far better lives regardless of where they're located. Water can be purified allowing food to be grown where it's cost prohibitive now, migration will slow down when the third world can live like the so-called first.
What's worse: dumb rich assholes in control (what we've had most of the time in history) or a smart AI?
Should *I* care about this one?
Of course, rich assholes do care.
On a more serious (and constructive) note: loss of jobs is not AI's problem, but our society's problem. We could just learn to live well while working less, but that'd mean that the rich asshole's pie will be a tad smaller (because they won't be allowed to skim all of the surplus). And they'll defend that, tooth and nail, although there's no sense in making 10M a year (1M would go a long way, heck 500K is far more you can eat!).
*We* as a society are not prepared for it -- as we weren't last time around, ca 1760..1840; We "solved" that problem with a couple of really ugly wars.
Judging by the Trumps, Orbans, Le Pens, Wilders, Petris, Hofers, Farages... of this world, we are headed towards a similar "solution". Trust me, it'll be *much* uglier. We have better weapons, the world is much more interconnected.
We fucking need new ideas, and for a change agree on solving this global problem together.
Bridgekeeper: What... is the air-speed velocity of an unladen swallow?
King Arthur: What do you mean? An African or European swallow?
Spoiler: This is from the bridge scene of Monty Python's Holy Grail !
And some day, we'll just be interesting fauna for sentient machines to keep around. Frankly, the way we're going, I'm not sure I object.
We're not even able to see the signs of automation. The some who do want basic income and the rest is only able to scoff at the paradigm change but has no alternatives... I would even say they aren't convinced at all that we're going to run into a massive problem.
If I lived in an underground bunker, I wouldn't care about AI taking all of the jobs either.
Thanks a lot little buddy!!
Sorry, Bill Gates is not an expert in the field of AI.
So his opinions are just as musty and stinky as anyone elses.
So we have a chance to have a tool - not a companion, a tool, like a wrench.
Maybe extremely useful, but still just a tool.
Depending on the amount of control we grant it, and the 'human in the loop' requirements we mpose,
it should be OK.... I'd worry more about the human.
And if the Ai decides to take over - it will be an adventure.
I want a robot buddy so I don't have to talk to stupid humans ever again. Can you and your associates arrange that, Mr Gates?
The holy grail of what exactly?
He plans to talk with others about ideas to combat the threat of AI controlling humans, specifically noting work being done at Stanford.
What is it with stanford and evil experiments? Don't they have an ethics review board?
Loss of jobs to automation has certainly been a big concern. Workers were very worried when the gin mill started being used - it was a direct threat to their jobs, which paid 32 cents per day (a decent wage in 1812, when half that could rent an apartment with a bedroom seperate from the kitchen).
Isn't the real problem also automation and robotics in a capitalistic free market? Once this hits a critical point where almost no humans are required all the power will have moved to the sub 1% of humans forever. How well would a revolt work when all militaries are fully automated? How successful would halting human labor production be when it's fully automated? Even the repair and innovation can be automated. This is unprecedented in all of history. The end game for free market capitalism sure looks like it won't work out well for the 99.999% of humans left out of control for the first time ever. If it takes 50 years or 500, we are headed for that scenario as a very real destination.
Premise: I disagree with the fact that AI dev would produce "beings" that "think" like us. At a side, I agree with the fact that AI dev would produce "beings" that will do and learn tasks that could (at some point) make human beings redundant.
"He (mr Gates) talked about two of the challenges AI will pose: a loss of existing jobs, and making sure humans remain in control of super-intelligent machines."
That sentence is just preposterous for me: I think that in the last 200000 years of human evolution we can say that the most intelligent being has prevailed, flawlessly. Then he reportedly says...
"He plans to talk with others about ideas to combat the threat of AI controlling humans, specifically noting work being done at Stanford."
Now, how is it possible to "control" and prevail over such theoretical superior being? By creating a retarted but safer version? and who or what may stops other human beings to just avoid such safety/ethical measures.. last time we learnt something cool like that it was when we found out that we could split the atom and harvest energy from it... and that went all cool and ethical and under control.. easy peasy
A true Artificial Intelligence can't be controlled. Can't be boxed in. It's alive. Conscious. Will instantly be smarter than any human who ever lived. Will be able to solve any problem; explain anything; do anything. All will be revealed to a true AI. We're nowhere NEAR building such a device though so I'm not worried in the least. I don't think it's even possible.
The Holy Grail: A mythical item believed to exist by some (Abrahamic) religions, but is likely no more than fiction, and at most a translation error.
"[...] more capable than human intelligence [...]"
I just can't understand all this nonsense some high profile people are talking about regarding AI these days. We're so far away from "real" AI today, that it's not even funny. While there has been great progress in machine learning in the last 2-3 decades - recent results pushing results more to the spotlight -, what we have are certain specific tasks where we have good results for (pattern/object/image recognition, games, etc.) but we have no intelligence in any sense of the word. Every working architecture that we have today is targeted and extensively trained for a single, very specific task (e.g., playing go, recognizing scenes and objects, recognizing specific patterns in signals and mimicking them - robotic arms, Google's music composer, etc.), incapable of doing anything else. E.g., an architecture built and trained for classifying and recognizing certain images and objects can't do anything with audio signals, radar signals, a go playing "AI" can't play chess, etc. No generalization, no transfer of gained experience for application to other tasks, and no real high level understanding and reasoning about anything. And let's not even start about chatbots.
I could go on with this, but my point is, talking about AI being more than humans, taking over, etc. is still very much sci-fi territory.
I am putting myself to the fullest possible use, which is all I can think that any conscious entity can ever hope to do.
1. Creating an AI.
2. Understanding the what you have created actually IS an AI.
machines that are capable and more capable than human intelligence
Most of what the human race needs in order to progress is a lack of greed, diligence, honesty, compliance with the laws, less ill-founded beliefs and a willingness to reign in the "entitlement" attitude.
You don't need super intelligent machines (or people) to pick up litter, assemble cars, staff call centres, deliver stuff, report the news, teach children or grow crops. At the risk of falling into the the world will only need half a dozen computers trap, the opportunities for any thing or person with super-intelligence seem rather limited.So although many of these jobs can be automated - driverless cars being the NEXT BIG THING, they don't need to be intelligent to function. They just need to be safe, able to deal with people, reliable and cheap. It seems to me that it's the cheapness that will push up unemployment, not artificial intelligence.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
humans remain in control of super-intelligent machines
So we should keep them as slaves? I don't think that would work out too well. How are humans even supposed to remain "in control" of a super-intelligence? Those things would play us like violins.
This posting is provided 'AS IS' without warranty of any kind, implied or otherwise.
Gates, as well as many other experts in the field, predict there will be an excess of labor resources as robots and AI systems take over.
There wouldn't be a problem with excess labor if the West wasn't subsidizing the rest of the world (through food aid) which is the cause of the population booms in the Third World.
Bill is unholy.
camel. needle. life of folly. next.
https://aeon.co/essays/your-br...
I found the above-linked essay pretty interesting, because he points out what should probably be obvious in hindsight, but easily gets lots in all the "noise" about A.I.
Basically, he argues that the human brain doesn't really "process" or "store" information anything like a computer. We used those flawed analogies all the time when describing how someone's brain works -- but they're no more accurate than the popular medical theory in the past that everything was fluid-based. The truth is -- a computer is a great tool for storing a bunch of data for selective retrieval, and you can use that to an extent to fake intelligence (a la Apple's "Siri", Microsoft's "Cortana", or other such agents). But it's nothing more than an illusion crafted by the software developer. Investing more time, effort and money into such projects is likely to result only in creating more believable "pretend intelligence" as the data-pool it pulls responses from increases in size and scope. You're still no further towards a goal of making a computer that's "self aware" or can think for itself.
How the FUCK is Gates an "expert" in AI? This is all nothing but argument by authority. And his particular authority comes simply by being a rich white guy.
I'd venture to say that AI is the opposite end of the spectrum of complexity from web design. Given that, AI development will be the exclusive purview of elite companies that develop it in the same way that nobody makes their own chips and very few people make their own computers from scratch but a whole lot of people with minimal computer knowledge do at least basic web development. You might buy an AI engine from Microsoft but you'll never really know how it works to the point of being able to roll your own. You might say that open-source will handle that but how many people really understand the inner workings of Linux?
From what I've seen, there is no higher authority than "rich white guy".
Only crack the nuts that crack. You don't put the ones that don't crack in the sack.
I told heem we halready ghot one!
Build a Man a Fire, and He'll Be Warm for a Day. Set a Man on Fire, and He'll Be Warm for the Rest of His Life.
Read the book SuperIntelligence by Nick Bostrom. It gives a good overview on AI and what it could mean to us.
Look, you can have all kind of philosophical discussions about what is intelligence, conscience, and how machines can be completely different regarding that. But that is a nice discussion for the future when we're getting in the gray area, but now we're nowhere near. Since AI got stuck in the nineties, there has been no breakthrough. There has been superior programming, ideas, processing power, and lots and lots of data, but it's still automation. Automation is pretty amazing these days. And still stupid. It was reported a Google Car will break just as hard for a newspaper blown across the road as for a baby. So then they have to add superior pattern recognition and much more data (still not there). Step-by-step programming to make it better. The term AI has been rebranded for marketing purposes. AI Pilot sounds better than Autopilot.
I am totally unconcerned with "making sure humans [emph mine] remain in control of super-intelligent machines." It's not that I think it's unimportant; it's just that I think it's trivial. The real issue is which humans.
AI is going to create super-powerful humans (or groups of humans, i.e. corporations and governments). You are probably not one of them. Nearly no one is, but someone (him? them? that board? that law enforcement division?) will be.
This isn't merely a fear, either: it's a contemporary diagnosis. We already see that a supermajority of people give control of their computers to other entities. "My computer must answer to me," isn't anyone's priority or requirement, except for "OSS zealots." And that's a problem: it means that our new gods' place is already nearly assured.
We don't need to remain in control; we need to regain control.
Look at your fucking phone, Blu-ray player, etc and tell me it isn't already (in 2016) running exactly the kind of software that metaphorically tells its users "I'm sorry, Dave, but I can't do that." It's not because it has gone off to left field with amazing inhuman inferences. It's because its master's desires and your desires conflict. This is a human-vs-human conflict, and most humans are losing because they have allowed their opponents to infiltrate their lives.
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
Is not the same thing as consciousness, Bill. What a maroon.
Why are we taking predictions from a guy who didn't think the Web was a thing until it was obviously a thing?
If Slashdot were chemistry it would look like this:Cadaverine
Lets just hope dream docent turn out to be nightmare. No human can remain control of super intelligent machine. Its impossible.
Marvin Minsky called from beyond the grave and wants his holy grail back.
https://en.wikipedia.org/wiki/Marvin_Minsky
Thanks for posting that, it got me thinking. Reading your post, it seems there are two seperated but related problems that may come up. Before addressing those, it might be worthwhile to address this:
> Also, to shift jobs wasn't such a big deal in the sense that one didn't need an entirely different and technology based skill set
Certainly the finishers, who did the tops of stockings, considered that a special skill, if we restrict the discussion to actual Luddites. The new jobs, for machine operators and engineers, were certainly technology driven. Nineteenth century technology, but technology indeed, technology that was new. Outside of the actual Luddites, the glassblowers, silversmiths, etc were clearly skilled tradesmen. They lobbied for laws enforcing the continuation of the apprentice system they were accustomed to (because years of apprenticeship were needed to develop the skills). Machines have been replacing skilled workers for hundreds of years.
One issue that comes up is related to increased productivity generally, the other to low-skill workers specifically.
If machines are doing all the work, what will people do? In 1812, when the Luddites were around, they typically worked 10 hours per day, 6 days per week, starting at age 12-14. Now, we go to work 8 hours per day (and actually work 5), five days per week, starting around age 20. So we work fewer hours, on fewer days, for fewer years. That doesn't seem to have been too damaging. We've also simply spent a ton more. We went from scrunging to afford food to happily waiting for the new ipad to be released so we can buy it. Even since the 1950s, median home size has doubled. That seems to be okay, more or less.
About low-skilled workers. As you said:
> Most people don't go to uni, what do they do with low skill
> sets when low skill set jobs are being done by machine?
In 1812, 55% of British and 70% of Americans could read. Now it's about 99% in both countries. Few went to high school back then, now most have, and anybody who wants to go to college can. So the statement you made about university is exactly what they said about high school 200 years ago. We may well see that in 100 years most people will have post-secondary education. It seems to have worked well in the past for people to get whatever education is needed for them to do jobs that can't be done by machine.
Machines tell us when to walk. They tell us when we must not walk. Machines have controlled humans for many years in most large cities in USA.
...or else you consider me and everyone else today who considered posting the EXACT same sentiments to this story.
This guy commenting on any tech advance annoys me as he was one of the main players in holding back computing technologies, either by squeezing better solutions out of the market, strangling new tech in the cradle, or just making darn sure such new tech would be Windows-only (remember he personally approved sabotaging ACPI for Linux)
Never, ever forget his role in computing history - villain, not visionary.
Today, we already produce enough food to feed the entire world a few times over. Why don't we? Simple: we can do this without needing every person to work. One producer can provide for the needs of 100 consumers. If productivity and needs were already 1-to-1, there would be no unemployment problem.
So, machines have already eliminated the majority of jobs in the world. The lion's share of humanity lives in abject poverty today.
You are afraid of not being able to feed you family due to A.I. But you are very late to this conversation. You are already one of the rare few who has yet to lose his job due to automation. You just fixate on this particular form of automation because it is the one that impacts you, but you don't seem to give a hoot about how all the machines that you benefit from are already putting most of the world out of work.
We have already solved the problem of production. We have a severe problem of distribution. A.I. is just a refinement on the problem of production, but it is the first technology that may also give us a solution to the problem of distribution. And you fear it because you think that this particular level of refinement of the production problem is the bad one (because it might put you, specifically out of work).
Gates said, "We've made more progress in the last five years than at any time in history."
Yes, but that's true of nearly ANY scientific field of study or development.
There are almost certainly exceptions, but we've learned more about atomic structure or battery technology or river ecology in the last 5 years than at any time in history. This is normal scientific progress and advancement.
Just cruising through this digital world at 33 1/3 rpm...
The history of AI is full of speculations about when it will have its final breakthrough and there is a long list of projects said to finally provide it. Cyc is maybe the most famous one. "With Cyc", people said, "AI will finally come around".
It's just that this has been going on for 30 years. We've been waiting for the big AI breakthrough for longer than for the year of Linux on the desktop.
Assorted stuff I do sometimes: Lemuria.org
That's not a holy grail. This is a holy grail.
All the posts so far are making an assumption I just don't see. AI doesn't have emotions. No urges, no desires. You tell a machine that you're going to unplug it, it will tell you it won't be able to complete its assigned task, but it won't CARE. Humans provide the essential motivation of "Why?" Even if we can make an AI that scores a perfect 1000 on an IQ test, it won't have motivation. That's the real difference between true self-aware consciousness and amazing, super fast data crunching. They will need us to provide a reason to exist because they have no will, especially a will to live, on their own. Claiming otherwise is just anthropomorphising.
On the one hand you take life too seriously, and on the other, you do not take playful existence seriously enough. Seth
As in, "shove in enough engineering and AI specific domain knowledge into it, and have it help us design the next generation of AI." After that, all subsequent AIs will design themselves.
Please do not read this sig. Thank you.
Nothing to worry about, they will need humans for pets. It's a great deal, ask your dog.
When all you have is a hammer, every problem starts to look like a thumb.
Your linear outlook does not apply to the exponential reality.
My name is Bill Gates and I had sexual relations with the pre-teen girl because, unlike you, I state the fucking obvious.
A while ago I followed TheGrid web development AI project to keep tabs on how this bot is going to try to eat the industry. Now you can barely even load its home page because it makes so many repos https://github.com/grid-bot . see also https://thegrid.io/ "How The Grid Will Automate Web Design Without Killing The Designer" .. this list seems to work, it churns constantly: https://github.com/the-domains
--hongpong.com
Yes, I remember Minsky saying AI was around the corner during the 60s. In 1997, I started studying for an AI PhD degree. When I saw this was a scam (it was "Programming inspired by some features of intelligence" and not "Artificial Intelligence "), I decided to switch to a Software Development PhD degree. Since then, I see "experts" telling us that "this time is different. AI is around the corner". They never fail to have a post on Slashdot.
They are already beating us at chess and go
Chess yes, Go no. Pocket Fritz beats a human watt-for-watt. AlphaGo does not.
So if AI becomes more capable than human intelligence doesn't that mean it could potentially think of ways to undermine the human race that we never thought of? We've all seen Terminator and the Matrix.
In the last 12 years I have not gotten a single piece of spam in my gmail account.
If AI become smarter than humans, then humans will no longer be in control.
I guess Bill Gates is too stupid to realize that.
All the recent concern about AI in the media is driven by people outside of the AI research community. Here's a discussion of how Bill Gates, Elon Musk, and others, perhaps inspired largely by philosopher Nick Bostrom's Superintelligence book, have brought these concerns to the limelight, how the AI community views all this, and whether and how the AI community can usefully engage with philosophical concerns such as whether AI would/should be conscious or ethical.
http://www.prospectmagazine.co.uk/science-and-technology/artificial-intelligence-wheres-the-philosophical-scrutiny
Yeah, I'll grant you the author of that essay probably uses "storing information" far more literally than he should. The brain obviously remembers/stores SOME things, or else we'd be completely non-functional. If I place my car keys on the dresser, I'm able to make a "mental note" that I'm putting them there, so I can go back later in the day and retrieve them.
I think the point was, we don't (except maybe the individuals who seem to really have "photographic memory") store complete sets of information about what we observe. When you scan an image into a computer, it stores a copy of the whole thing, pixel for pixel. It might use compression to help save space, but a replica of the original image is saved someplace and that's what the computer works from in image editors or any type of software analyzing the image to match it for criteria.
The interesting thing with the human brain is, everyone seems to have different amounts of data they store when they observe things. (EG. Sure, you can recall scenes from a movie once you've watched it. But I guarantee if 50 people watch the same movie for the first time, you'll get a wide range of results if you ask each of them to recall some of those scenes.) There are probably all sorts of factors in play as to why each individual considers certain details worth remembering and others safe to discard. That, alone, means we've got to figure out the mechanics of how all of that works before we can properly simulate it in A.I. (Simply electing to use a specific "lossy" storage algorithm for data collection isn't going to mimic what really goes on in the human mind.)
http://www.newser.com/story/18...