Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Eagleon]

I feel like a lot of people just view intelligence as a stat and think we're going to make a robot with a higher number in that stat than us someday without thinking about what it entails. What makes an intelligence greater than ours and not just different?

What problems would a higher intelligence solve? I can't think of any problems that exist because humans lack the intellectual capacity to comprehend solutions. Maybe some math and physics stuff, but nothing social.

Would people even listen to it? If it offered a solution someone didn't like, I feel like they'd just accuse it of being biased. And they would have a point - a smart person can be more biased than a dumb person, and this can cause smart people to say very dumb things. How do we tell if it's actually getting better answers?

The outcome I would love to see, just from the point of pure humor, is that we make an AI and it becomes absolutely obsessed with making video games and writing poetry or something. It's really, really good at it too. Ask it to design some mining equipment, it shrugs its shoulders and says "I'm no good at that." Everyone gets fired, AI company goes under in massive investor shitstorm, and we refocus all of our energy into AI developmental psychology to make the best AI parents and schools and improve the minimum AI wage and subsidize AI colleges so that they can skim and wrangle their way into being the same shitlords as regular schools, etc. etc. The Future is Stupid.

>  raw computational power, which judging by our current obsession with mobile devices isn't even a market imperative

This is a silly statement. Total world computation power isn't tied to the amount of CPU power in your pocket. Anyway, mobile device makers cram as much power in their as they can. People certainly pay a lot more for a powerful phone than a basic one right?

Anyway, mobile computing really doesn't have anything to do with the topic, so it's kind of questionable why you'd even bring that up.

They're usually paying more for a bigger screen. I bring it up because it highlights that Moores law depends on us valuing making computers more powerful, which was true when they were kind of shitty. After a certain point, there may be a point where it's just shitty enough, and incremental or even just fashion-wise improvements (COOL! Our displays are thin and totally clearseethrough now! It's the future!) become acceptable to the general public.

[Reelya]

hat problems would a higher intelligence solve? I can't think of any problems that exist because humans lack the intellectual capacity to comprehend solutions. Maybe some math and physics stuff, but nothing social.

I'm actually not to sure about that. Optimizing things that have more than a handful of variables has never turned out to be something humans are specifically good at.

Consider the Eurisko example. Many smart humans spend years fine-tuning fictional ship loadouts in a competitive game environment. Then a dumb program comes along and runs for a few months in the spare time on someones PC and designs a whole new way of playing that blows every human out of the water. Not by a bit, but by miles. Humans pretty much suck at optimizing anything but the most simplistic scenarios. There are definite social gains to be had by using AI to analyse data.

The machine doesn't even have to come up with solutions that are perfect. Just better than what we do. That's a lot easier than it sounds since most of our solutions suck and we often ignore the data in favor of ideology.

I think a lot of the problems solved will be the types of hideous inefficiency and waste where you only realize how bad they were once the new way of doing things is implemented: those types of innovation that were fought tooth and nail until they came in then everyone quickly jumps on board, because they are so clearly superior that instantly the old way of doing things seems so shitty that you would never go back to it. These kinds of things happen all the time. And it's where computation can play a big part. Often a solution is not implemented because it's not intuitive enough, even though all the data says it's a winner. Limited human perspective has never worked great for complex systems.

Hell, probability experts were all stumped by the Monty Hall problem, and that is one of the most basic mathematical truths you can put together. If world experts in probability couldn't even comprehend very simple probability systems once you throw some "unexpected" element in, how do you think humans do in general designing systems involving millions of people?

[FArgHalfnr]

So... We know what the problem is: Global warming. We know what causes it: Greenhouse gas. We know how to fix the problem: move away from fossil fuels. We also know what prevent us from fixing the problem: market interests. Now, how to we overcome this? There must be a way to make the protection of our species and environment more important than short term profit.






...That last sentence made me depressed.

[Frumple]

If we consider the moment the first computer was created (A quick google search tells me it was built in 1822) as the moment we started working on a better version of ourselves, that's almost 200 years of development, and we aren't even close yet.

Do people really not really grok how freakishly fast these developments have occurred, compared to general human history? We speak of two centuries like it's this some strange long time, when in fact it's been a period of development more rapid than any before it by several orders of magnitude.

We've seen ridiculous transformations in the breadth and depth of technology in single lifetimes, and we don't actually have much indication that's going to slow down. Prediction is fairly silly, though, if only 'cause we don't really have a bloody clue what's going to change next. What has been happening over the last handful or two of decades is entirely unprecedented in human history. We don't really have a goddamn idea what's going to happen next.

On the other hand, to think that it's not going to perception shattering for like the sixth bloody time in the last century or so is kinda' foolish. We have had a string of successive events telling us that predicting things aren't going to radically change is somewhat unlikely, with every indication that's either not going to slow down significantly or begin to outright accelerate.

Maybe it'll be AI. Maybe it'll be something else -- we're seeing gains towards biological freaking immortality right now, nevermind the various non-medical advances. The only bet you can really make at this point is that it's going to be something, because that keeps happening.

[WillowLuman]

Don't we have a thread about the singularity?

[SirQuiamus]

Consider the Eurisko example. Many smart humans spend years fine-tuning fictional ship loadouts in a competitive game environment. Then a dumb program comes along and runs for a few months in the spare time on someones PC and designs a whole new way of playing that blows every human out of the water. Not by a bit, but by miles. Humans pretty much suck at optimizing anything but the most simplistic scenarios. There are definite social gains to be had by using AI to analyse data.

I don't find that very convincing. We should keep in mind that it was a competitive game, not a collective effort of problem-solving. The smartest players were all tinkering with their own ship designs without sharing their knowhow with others – the exact opposite of what scientists are supposed to be doing with their research. If thousands of players had purposefully co-operated together to create the most efficient tactics, would the end-result have been qualitatively different from what Eurisko came up with? And what if every player had access to a personal computer for testing and optimizing their algorithms with the help of relatively simple programs? And what if each of them had a doctorate in mathematics? I don't see any reason to assume that a transition from a team of thousand super-intelligent scientists to a super-intelligent AI would mark a giant qualitative leap in intelligence.

Also, this:

We know what the problem is: Global warming. We know what causes it: Greenhouse gas. We know how to fix the problem: move away from fossil fuels.

     

[Fniff]

I wonder how everyone would react if we made a superintelligent AI, and told it to come up with a solution to climate change/social issues/war/etc...
And a message came up saying "ERROR: SOLUTION NON-EXISTENT".

[Angle]

So... We know what the problem is: Global warming. We know what causes it: Greenhouse gas. We know how to fix the problem: move away from fossil fuels. We also know what prevent us from fixing the problem: market interests. Now, how to we overcome this? There must be a way to make the protection of our species and environment more important than short term profit.

Not just market interests, there's a healthy dose of human stupidity - "I refuse to admit this is happening, or even consider the evidence, and even if it is happening, I refuse to do even so little as send my politicians a letter over it." We're getting better about it, but it's precious slow. The internet seems to be speeding it up a little, but people are still close minded foolish creatures.

[Reelya]

Wars for food are coming unless we figure out an alternative ASAP.

recycled sewage sausages. Also, insects and other sources of nutrition. This stuff may be unmarketable now, but try mass starvation and see how it goes. Food supplies aren't the problem, humans being fussy eaters is.

I think we should really be investing much more into fish farming. It's not like there's a shortage of the raw material - salt water - to make this happen. And you're not dependent on rainfall or soil conditions like you are for land farming, just maintain the salinity and chemical balance of your breeding tanks. In the long run, fish farming could be more economically viable than sea fishing. Consider the economic scales of animal farming vs trying to source meat from hunting. Domesticated fish in farms would probably undergo changes similar to domesticated land animals, and we could selectively breed for yield and growth.

[Frumple]

I don't see any reason to assume that there would be a giant qualitative leap from a team of thousand super-intelligent scientists to a super-intelligent AI.

... have you ever seen a team of a thousand scientists actually work together well? There's significantly fewer logistic issues involved with one 1k scientist equivalent computer and a thousand scientists. You'd see a significant qualitative leap simply because there's significantly fewer human inefficiencies involved.

In other words, I see a thousand reasons why there would be a giant qualitative leap from that team to the AI (or expert systems, or whatever, really). One for every scientist in the metaphorical team. Probably more just due to the mechanics of communication and cooperation and whatnot. You get some gains back from synergy and whatnot, but I'd be incredibly surprised if it was enough to offset the difference.

The reality, of course, is that you'd use both to the extent you're able. S'already what we do to a fair extent, and there's no real reason to stop until the scientists are well and truly made obsolete.

[Reelya]

Humans also have cognitive blind spots, even the best of us. There are a lot of things that trip humans up, but computers just see right through them. Humans and AI are also nothing alike. It's not a quantitative thing at all. It's definitely qualitative, in that an AI with the same effective computing power as a room full of humans will give you totally different answers. Stuff you never expected.

A lot of things designed by neural networks or genetic algorithms look weird, alienesque, in a way no human would have ever designed stuff. Non-symmetrical and freakish. But they work much better than the hand-designed ones. Take a look at NASA's super-antenna designed by AI:

http://www.nasa.gov/centers/ames/research/exploringtheuniverse/borg.html

No human would ever have designed it like this. A room full of antenna experts would never have produced this design. But it's by far the best design.

[WillowLuman]

Inb4 derail into the singularity argument about whether humanity can be rendered obsolete.

[SirQuiamus]

Humans also have cognitive blind spots, even the best of us. There are a lot of things that trip humans up, but computers just see right through them.

My point is rather that if each scientist has a computer with sophisticated software, there's no need for a hypersmart, godlike Multivac. We have to rely on technology in all things, especially in what comes to mitigating the effects of human error, but we don't need an artificial saviour to tell us that the apparent solution is truly right.

[Reelya]

I think that's a strawman right there. The concept of a singularity in computing power and design doesn't require any of the things you complain about.

Also the perspective that "if each scientist has a computer than can do X it's good enough" fails to realize that it's our existing social structures that may be holding us back. It's kind of like saying if each carpenter in France had a better set of tools we could totally build the Eiffel Tower. Filtering anything through the limited perspective of individuals is going to miss out most benefits that come from synergy.

Look at things like that antenna in my last post, which is a super-antenna able to communicate between satellites and earth for a fraction of the energy and cost of a human-designed one. And completely incomprehensible to a human as a design. Then consider that the same sort of programs could be applied to architecture, public transport and traffic systems, the financial system. It's pretty certain that there are many areas of human social endeavour that are not optimal, and the optimal solution is utterly counter-intuitive in the same sense as space antenna design is counter-intuitive. But computational techniques can find those solutions.

[FArgHalfnr]

So... We know what the problem is: Global warming. We know what causes it: Greenhouse gas. We know how to fix the problem: move away from fossil fuels. We also know what prevent us from fixing the problem: market interests. Now, how to we overcome this? There must be a way to make the protection of our species and environment more important than short term profit.

Not just market interests, there's a healthy dose of human stupidity - "I refuse to admit this is happening, or even consider the evidence, and even if it is happening, I refuse to do even so little as send my politicians a letter over it." We're getting better about it, but it's precious slow. The internet seems to be speeding it up a little, but people are still close minded foolish creatures.

In the past, changes of opinions like that would work on several generation, as each new one was more used to seeing whatever was new than the last one. We know that we don't have the time to wait for the previous generations to die, so we must find a way to change the opinion of the population. If we look in the past, we can see a few cases where there was a relatively quick change of general opinion, such as what we see in periods of war when people are influenced by massive propaganda campaigns. So that seems to be our best bet. The internet can be a good tool to spread the message, but I suspect that a good chunk of those who are in denial don't use it. We'd need to gain access to other medias, such as the television and newspapers, but those are owned by private interests, so we need to make it so that they believe that being in favor of global warming denial is less profitable than the opposite.