What do you think? Is he right? Should we really be so conce

>>8423249
Pseudo-intellectual bullshit, moving on.

>doesn't work in AI
That's like a lawyer telling a doctor how to do his job. Learn how to stay in your expertise.

I watched the entire and I was left with the impression "what is intelligence." I realize that's not the sort of impression that will be popular on sci, but nonetheless it was hard to decide whether I agreed with his portent of doom since I couldn't decide how plausible it was without knowing better exactly what it is he's so certain we could create. I personally don't believe AI can be created but then again my justification for that is Godel incompleteness theorems. Still I'm no expert so I don't regard my opinion as useful.

Why is contemplating the ramifications of AI pseudo intellectual? That means it isn't actually an intelligent pursuit to inquire about the effect an invention could have? So what's a better use of fifteen minutes? FB and video games? Some of the most intelligent men in history created the atomic bomb and spent decades afterwards agonizing over its implications. I would think, for example, Richard Feynman's intelligence favorably compares to the typical, ridiculously self assured sci posters. You know I think some people take scorn of reflection far too seriously as a sign of "real" smarts. It might surprise some of you to discover smart people actually think about stuff.

>>8423348
We've already created AI. It's been in video games for decades.

Here's an actual expert on the danger of AI: http://www.youtube.com/watch?v=tcdVC4e6EV4

Stop listening to non-experts, dumbfucks.

>>8423313
It's posts like these that remind me why STEM nerds need to start taking philosophy more seriously, especially epistemology. The great scientists of the first half of the twentieth century were all very philosophically literate. At some point that changed, and for the worse.

>>8423395
I don't disagree but by the same token the philosophers should start becoming scientifically literate. Descartes' knowledge of math and science probably eclipses that of most philosophy professors, even though he lived hundreds of years ago.

>>8423249
Non-mechanical thinkers suck. I don't mean like material science, gears, or mechanical systems, I mean people who can't generate a realistic spectrum of possibilities, see how they work and interrelate, then in serial weight them by probability and evaluate them relative to how the world really works. This is basic prediction, and I just refuse to believe people who get it so wrong are capable of actually executing this process correctly.

The dangers of AI are obvious when you stop thinking of life as being unique. We're machines. Scale is arbitrary, delineation between systems is arbitrary, macro and micro is arbitrary. That's what reductionism is, and that's why reductionism works. So use it. And use it right.

Things simply act in accordance with their nature. They're machines inside a bigger machine. Your everyday role in life is possible not meaningfully different than the thermocoupling in your stove controlling when current flows. Put AI in this same heavily abstracted context and realize the capabilities it has. Humans are stupid. They will with certainty network it, or at least provide it with inputs. There will be holes in a system that it may exploit to gain more physical control, if, that is in accordance with its nature. And given that Google, Facebook, Microsoft, etc, are more or less trying to create God, of course it's going to be networked.

I don't even see why this is a question. It answers itself.

>>8423368
That's hardly AI. It's a simple system capable of adjusting some variables in a very constrained system.
Conceptually it's like a simple algorithm which deduces the parameters in y(x)=ax+b from a graph with well-defined limits to what it can do.


Honestly walking/generating trees and graphs ain't exactly a learning capable self-aware intelligence.

>>8423444
You know SciFi isn't reality? Read some textbooks and lay off the Marvel movies, friend.

>>8423431
>I don't even see why this is a question. It answers itself.
If the question is a simple yes or no with absolutely zero context about a theoretical possibility then yes, probably there will be some kind of smart AI which can fuck us in the ass.

However you're exactly incapable of doing what you want others to do.
The complexity of working life and the intelligence it produced is enormous. Our understanding of it for practical, let's make a better race than us or simply an intelligent race tier stuff is zero.
For many problems we have experimentally verified solutions and we know how lot of systems work at a higher abstraction level, but the deep shit is so far away that we don't even have a guess how it will look like. See current physics and how enormous it is, the predictions scientists are verifying and throwing out all day.

The AIs we make if you can call it that aren't really AIs. They basically have an algorithm which makes them produce a specialized solution by solving a generalized problem using their built-in algorithm written by humans.
A PID or Fuzzy control isn't exactly AI in the sense most people use it.
Other approaches which try to model biological systems are still in their infancy. Even artificial neurons have models fucking far away from real biological systems most of the time.
We haven't even considered emergent behavior and whether it is necessary to have the same kind of emergent behavior in different systems.

>>8423444
>That's hardly AI. It's a simple system capable of adjusting some variables in a very constrained system.
In the same sense monkeys are intelligent and conscious animals that is, in fact, real AI. The fact that it's primitive and undeveloped doesn't change that fact.

Your goal posts are way too high, btw, not everything that isn't artificial consciousness is primitive or worthless. The space between most basic rudimentary AI and true artificial consciousness is so vast as to be nearly endless. At the higher end of the scale there is some extremely intelligent AI that could replace most jobs and that could pass Turing test but that still isn't conscious.

Why not program the AI with a automatic kill switch if they harm humans?

>>8423467
Or just Asimov's three laws, taking into account loop holes he showed existed.

Also people get far too much of their knowledge/understanding of AI from films and films always follow the tired "deadly killing computer that wants to kill humans" trope.

If we're actually capable of creating conscious AI we'll also be able to create its personality and quirks that go along with it, you will literally be able to create hyper active always happy anime girl AI if you want to.

>>8423467
Why not program cells with a set of pathways that act as automatic kill switches if they're apt to harm the organism as a whole? Wait...

>>8423457
Are you sure want to tell that to me and not the other guy with video games?

>>8423460
You're right, however I don't think most people think about that kind of AI when they talk about AI. They most of the time mean some popsci tier stuff (muh the Terminator is here... asdasdasd).

For example as far as I know (I'm no biologist), but there are (certain? or maybe all idk) insects which technically still try to do their stuff if half of their body is missing. They have no concept of self like us, yet they can solve enough problems to be self-sufficient at the level they need to be.

File: 1458319512549.jpg (151KB, 820x545px) Image search: [Google]

151KB, 820x545px

>>8423470
>You can create your happy anime girl AI
Humanity can do so much right in this world it's great.

>>8423467
>Asimov's three laws
Don't cite these in discussions about AI please, they're completely unfeasible bullshit

https://www.youtube.com/watch?v=7PKx3kS7f4A

>>8423483
>Humanity can do so much right in this world it's great.
First robot I thought of when considering robots with autonomous, human like bodies is cute hugging robot in the shape of a (not too young) girl. Wouldn't it be amazing to come home after long day and be greeted by happy smiles and hugs from your cute robot?

And since it's AI it can be programmed to be permanently at the level of happiness that would be psychotic in real humans and with just enough but not too much intelligence.

>>8423492
>in the shape of a (not too young) girl
That's called a woman.

Also, do you mean happy hugging woman who look cute and sexy and you can fuck after a long day to relieve sex?
I'm all in. Where to finance the kickstarter?

>>8423500
>relieve sex?
I meant tension. fuck.

>>8423484
meant to quote>>8423470

>>8423459
>If the question is a simple yes or no with absolutely zero context
You can't have "zero context". Humans have already thoroughly investigated the universe, and it's made it very, very clear what macro machinery it's capable of supporting and what unavoidable barriers are at play. AI is obviously a real possibility. There isn't so much of a "theoretical" element as you seem to believe. It's just a matter of how, not if. How. It's also a matter of architecture. The current hardware we're using is not conducive to AI. Specialist hardware is now shaking out how to balance distributed and centralized processing capacity of a given node.

Anyway. A general intelligence doesn't need to model us or our way of thinking to be net superior in every conceivable way. I have more background in biology than computer science (though I know C++ and can program decently), so I'm approaching this from a different angle. The goal is not to make an artificial brain, it's to harvest the underlying mechanisms for performing a given micro, or macro, operation, and apply them elsewhere.

>it's a dude with absolutely no technological knowledge attempts to speak as an authority on a technical subject episode

>>8423500
>I'm all in. Where to finance the kickstarter?
AI wise we could probably already do it. The issue is the body. Even the most advanced robotic bodies we have right now are, to put it mildly, shit. They're too stiff and they have extremely limited range of motions and they can't even replicate basic human actions like simple, not genki girl smile,

The brunt of progress will have to happen in robotic bodies and cybernetics first before we have anything like that.