After the A.I revolution, the only jobs left on the market are going to be for arts and humanities.
How does this makes you feel /g/ ? STEM grads are working to kill their own jobs
stop conflating code monkeys (the cashiers/line cooks of the software industry) to people who actually design the software/algorithms/computers/etc
>>53612567
There are already computers that can produce art and music but still none that can write computer programs. Basically you're dumb.
>>53612567
http://www.bbc.co.uk/news/technology-34066941
>Midwife
I'll be fine.
>>53612567
I can think of supply chain managers and data analysts getting wreck by AI.
>>53612567
There won't be any general intelligence machine anywhere in our lifetimes, maybe never.
The combined computers, network relays and so on on earth are not even close in number with how many neurons a brain dead person has.
Sure specialized self learning software that will be implemented in lots of domains will boost progress but nothing more.
Your fear of AI is just the result of consuming hollywood media.
+ all this meme is just the result of esoterism clashing with science - and the mystification of it by normies that lecture popscience.
+ all the popularity hinduism got in America, leads to fucked up interpretations in everything.
I think an average American is having a more hinduistic mindset that he thinks he does:
Let's do some yoga for destressing ourself.
Meditation helps you with work.
Click this article it will explain you why your life sucks - because of past life mistakes.
Magic is real yo, all religions are right yo.
>>53612852
>The combined computers, network relays and so on on earth are not even close in number with how many neurons a brain dead person has.
True. But a computer is ridiculously more powerful than a neuron.
>>53613006
Not really
>>53612567
I'll just grow my own food in the garden.
>>53612605
This.
Most coders I know couldn't design their way out of a paper bag. But wait, I can hear you crying, "I write my own software all the time." Yeah, using established design principles.
Whatever happened to the idea that we can get machines to do our work for us so we don't have to?
Why is it so important that everybody has to work 8 hours a day? It's like a religion at this point.
I think the real question is: when can I get my mail-order Alicia Vikander?
>>53613163
I guess we dont like to feel like sheeps in the barn.
>>53613087
ha.
>>53613295
hue
>>53612852
>There won't be any general intelligence machine anywhere in our lifetimes, maybe never.
People saying this really pisses me off because such an advancement isn't as much subject to timing as it is inspiration. You can't count on breakthroughs happening or not happening with things like that. General AI could literally be cracked in 4 years if some genius suddenly figured out an algorithmic theory of consciousness. It could also take 6 million years. There's literally no telling the time on something like this and saying that we won't see it in our lifetimes is about as valid as saying we will.
>>53612567
>based that cuck's porn history on that robot
that movie was basically some smart guy trolling some faggot from /g/ when the vir/g/in thinks he is outsmarting the owner of the lab, he ends up dying of starvation.
good job /g/
>>53613087
Are you stupid?
>>53613163
Literally just dumb/untalented people don't want to feel useless. I had a talk about this with a friend of mine. He doesn't want machines to get rid of all the menial labor because people like him wouldn't have a reason to exist anymore.
>>53613321
You can't dream about walking trough walls because that is impossible, deal with reality.
>>53613382
What does that even have to do with what I said?
>>53613372
Educate yourself, neuron acts as the central core of the individual computing module - capable of so many complex computations and understanding them on its own - the dendrites of the neuron are capable of creating electrical spikes on its own but also capable of understanding the ones it receives - also add the synapes around the indiviudal neurons which server are complex memory banks and information patterns which are accessed at ultra fast speeds - not connect all the 100 billion neurons with roughly 1 quadrillion 1 million billion connections known as synapses.
Brain regions are specialized and optimized and there are many types of neurons.
For now our tools and technique of researching the brain are primitive and such the state of current understanding so there is not a proper way to measure exactly how powerful an individual neuron is - but clearly computer metrics do not translate into biology.
>>53613411
No amount of inspiration and will to achieve it will make you go trough walls, so both time and of inspiration are irrelevant.
Same goes for achieving synthetic consciousness.
>>53612852
Not necessarily true. Moore's law puts a CPU with the computational power of the human brain somewhere around 2030. While that isn't a law in the strictest sense, there is little reason to believe the trend will end anytime soon.
Now can the CPU make as efficient use of that power as our brain does? Probably not, but if nothing else, by 2030 we will be on the cusp of the brain in a microchip.
As for software, I think neural nets are looking very promising in their abilities. I don't think they are the answer, but I can image a sort of net of neural nets (neural-internet) running on those very power machines. It is easy to imagine them having some sort of emergent behavior that blurs the line between advanced program and sentient.
Based on the demographics of 4chan, most of us won't start dropping in large numbers until 2050-2060 so I think most of us will witness intelligent AI. Maybe not general intelligence in the strictest sense, but smart enough that it is hard to tell the difference. And at that point, what really is the difference?
>>53613488
We literally cannot say whether or not synthetic consciousness is possible because we don't even understand the nature of consciousness yet. Your post is a prime example of exactly what I was talking about before.
>>53613498
>Moore's law
Yes add concepts over concepts and keep dreaming on.
>>53613468
>accessed at ultra fast speeds
200 miles an hour is not "ultra fast". We have cars that move faster than that. And most electro-chemical reactions are actually a lot fucking slower than 200 MPH, only the very fastest reactions are that fast.
As opposed to moving electrons along a copper wire, which happens at the speed of light. How fast is that again? Oh yeah, 186E3 miles per SECOND.
Get the fuck outta here, lad.
>>53613468
Electrical computers can process and handle more data than a biological one which doesn't store most of the information its sensors pick up and tends to use shortcuts for compression purposes that are radically error prone.
>>53613572
go to japan and ride a train that goes 603 km/h (375 mph) on electricity
its not about the reactions or whatever you said
its about how you implement them and use them for your advantage
>>53613631
there's no such thing as a biological computer.
Brain works in different fashion, can't compare it with computer logic.
But anyway this discussion is pointless, non-productive and AI is a waste of time and resources. Enjoy your memes.
>>53612635
>There are already computers that can produce art and music
Top fucking kek
>>53613651
>AI is a waste of time and resources.
LEAST INTELLIGENT COMMENT OF ALL TIME
>>53612852
>he thinks 1 computer = 1 neuron
>>53613685
What's funny about that?
>>53612567
You're missing the point. There won't be any need for a traditional job in a post-work society.
Basic income will be able to be supplemented by additional work, but it certainly won't be a requirement.
People will be able to pursue arts and humanities purely for self interest.
>>53613711
*neurone
>>53613651
>Brain works in different fashion, can't compare it with computer logic.
If we're going by the accuracy and volume of data processed we don't even need comparable units to know that the human brain can't stack up to a computer in those regards.
>>53612567
>After the A.I revolution, the only jobs left on the market are going to be for arts and humanities.
The results of the forefront of neural net research excel primarily in art.
Computers have been able to synthesize decent music for years and only get better.
>>53613756
and eventually humanities will be obsolete because there will be no humans around.
>>53613468
is this bait? neurones function is about as complex as copper wires not full on computers you autist, they're capable of passing action potentials down their axon and releasing neurotransmitters, that's literally it
>>53613820
>>53612635
what is your definition of art?
>>53613651
>AI is a waste of time and resources
>Self Driving cars are a waste of time and resources
>Automation is a waste of time and resources
>Pilot assistance for aircraft is a waste of time and resources
>Data Mining is a waste of time and resources
>Automatic image classification is a waste of time and resources
>Automatic threat monitoring is a waste of time and resources
>Personal digital assistants are a waste of time and resources
I feel like the only impression you have of what AI is has been taken from hollywood movies.
>>53613938
>>Data Mining is a waste of time and resources
Not him, but I'd agree with this one to the extent that the resulting data is used to sell people yet more shit.
>>53613974
thats what facebook and google uses it to. But Is used a lot in different fields.
>>53613498
>Moore's law
That law has been proven false 10+ years ago, buddy. Move on to new memes.
>>53614347
this is the true nature of exponential grow. And moore law is just another iteration of a logistic curve. After a slow down on silicon chips there will be another paradigm change (like graphene, or diamonds, or another shit) that will keep up the exponential curve.
>>53615105
forgot the image...
>>53612567
>arts
>market
>>53613904
>what is your definition of art?
Computer programs exists which can generate classical music in a style of a given composer, indistinguishable from the real thing by an expert panel.
If this is not a good enough definition, I don't know what is.
>>53615184
>expert panel
Kek, same type of people who jerk off about drawings of monkeys.
>>53612567
>STEM grads are working to kill their own jobs
that's the definition of a useful job
>>53612567
retard detected. if A.I. is capable of enginuity it will also be more capable than humans at the arts.
>>53613544
Moore's law has literally been 100% true so far so shut up faggot
singularity fags are worse than religious people
>muh singularity
These people are basically a cult. They don't know anything about AI. Everything they believe comes from shitty pop media like The Matrix and Buzzfeed articles, not science and research. They have never read a paper on neuroscience or AI, much less have a doctorate. We should just gas them all.
>>53612567
i dont care op, killing all jobs seems like a cool thing to do.
>>53615462
>>muh singularity
>These people are basically a cult. They don't know anything about AI.
>Kurzweil is a nobody
>Hence why all the "Alphabet" op
>Hence why the calico funds
etc
>>53612567
>How does this makes you feel /g/ ? STEM grads are working to kill their own jobs
No, we are working on the machines that will finally build an Utopia for mankind.
>>53612852
>Your fear of AI is just the result of consuming hollywood media.
>+ all this meme is just the result of esoterism clashing with science - and the mystification of it by normies that lecture popscience.
>+ all the popularity hinduism got in America, leads to fucked up interpretations in everything.
Agree.
>There won't be any general intelligence machine anywhere in our lifetimes, maybe never.
Intelligence may be general, but nor capable enough to solve any human-level task. For example deepmind's http://arxiv.org/abs/1602.01783 is quite general, it solves several completely different tasks. But it's far from being human-level yet.
>Sure specialized self learning software that will be implemented in lots of domains will boost progress but nothing more.
I don't agree about "nothing more". A subhuman but general AI can be trained once to do empirical science in domain of your choice. Then you run 1000 copied instances of said AI in AWS or GCP, and they solve your problems for you.
Effects on human wealth can be profound. Antiageing research is the obvious potential application (Google/Deepmind's investments and rhetoric confirm this).
>>53613321
>consciousness
C-word is a meme. The best measure of AI agent's generality is its total score on a benchmark composed of various environments. You don't need any "consciousness" in an AI system, you need "only" learning & general problem solving capabilities.
>>53615723
fucking recapcha getting more annoying everytime
Lol @ "AGI is not within our lifetimes". Most of the software issues were solved within the period between 1950 and 1990. All that's left is hardware.
Disagree? Well, then you're disagreeing with Demis Hassabis, Shane Legg, Geoff Hinton, Rich Sutton, Juergen Schmidhuber, Bart Selman, Dileep George, Yoshua Bengio, etc. etc.
Even 30 years is becoming conservative at this point. Expect human-level AI 'any day now', frankly. My estimate is roughly 2030... and has been since 2005.
This is a dangerous game and we need to solve the friendly AI puzzle immediately. "Human-level AI is hundreds of years away, or maybe impossible!" is an absolute joke to anyone who's been following recent progress. Get real. You're literally endangering our species by ignoring risk.
>>53615723
>Which man is dressed more flamboyantly
lol
>>53615799
Literally who are those people? What universities were their PhDs in AI from? What advancements have they msde to the field?
>>53615799
>Get real. You're literally endangering our species by ignoring risk.
LW shill detected (^:
This is true though
>Disagree? Well, then you're disagreeing with Demis Hassabis, Shane Legg, Geoff Hinton, Rich Sutton, Juergen Schmidhuber, Bart Selman, Dileep George, Yoshua Bengio, etc. etc.
Staggering progress has been made since 2010.
Still, machine learning AI is very stochastic, it learns by mistakes, and learns realtively slowly. I highly doubt that it can immediately start doing something unexpected.
> solve the friendly AI puzzle immediately.
There are some solutions that look promising http://www.fhi.ox.ac.uk/wp-content/uploads/MDL-Intelligence-Distillation-for-safe-superintelligent-problem-solving1.pdf
Also Oracle AIs.
>>53615832
>>53613488
I can actually walk through walls
Checkmate atheists
>>53615843
If you don't recognize those names, I can safely assume your entire knowledge of 21st century AI has been crafted by Michio Kaku's "retarded cockroach" analogy. Do some research familam...
>>53615863
I'm not afraid of AI with 'volition'. Paperclip scenarios scare me much more.
>>53615843
Can't you even google anon?
> Demis Hassabis
CEO of deepmind Phd University College London
>Shane Legg
co-founder of deepmind, Phd
>Hinton
famous deep-learning researcher, Phd, works at google
>Rich Sutton
Literally top researcher in Reinforcement Learning, Phd UC Alberta
>Schmidthuber
very famous AI researcher from germany, Phd
If you don't know these names you are a meme.
>>53612852
Exactly this. What we call "Artificial Intelligence" is basically just fancy machine learning and smart brute forcing.
We're still waiting for the Newton of Artificial Intelligence just to come up with a basic framework. Then an Einstein to take it to the next level.
Making the kind of robot in Ex Machina is literally thousands of years away.
>>53613321
It doesn't work like that. Breakthroughs come after a certain prerequisite threshold of knowledge has been reached. That's why calculus was invented independently around the same time. It's not like someone sitting around will suddenly come up with a theory to explain intelligence. The person who does this centuries from now will be well read in every major AI paper leading to that point.
Humans can only reach the edge of human knowledge and push it ever so slightly forward. Once in a blue moon doing so pushes it over a hill and we get runaway breakthroughs like Newton and Einstein.
>>53613685
He's right, tiger. Do your research, sport.
http://www.bbc.com/future/story/20140808-music-like-never-heard-before
>>53615916
>I'm not afraid of AI with 'volition'. Paperclip scenarios scare me much more.
I understand, I have read Bostrom etc etc.
My point is that AGI will be a descendant of modern ML algorithms (in particular deep learning), and modern ML is known to learn slowly, continuously, udner constant supervision, and while doing many mistakes.
Such AI looks different from superintelligences that bostrom describes (bostrom's SI are "self-impproving", capable of designing complex plans and implementing them without single mistake etc).
Also I don't believe in strong "recursive self-improvement" because it has never been demonstrated (if you don't count very limited genetic programming experiments). ML models become better/smarter only when you train them on data.
>>53612567
No jobs? Fucking finally, where do I sign up?
>>53615988
I think you'll be surprised (maybe disturbed) to find out how far statistical modeling and traditional neural networks take us by 2030.
>>53615988
>is basically just fancy machine learning and smart brute forcing.
If its enough to build an utopia, why care?
>Making the kind of robot in Ex Machina is literally thousands of years away.
>Conflating robotics and AI
If you have a good anthropomorphic robot (a really hard feat, not yet achieved), it will be fairly easy to emulate the behavior of your average woman. Women are not smart, anon (^:
>We're still waiting for the Newton of Artificial Intelligence just to come up with a basic framework.
>Haven't heard about Marcus Hutter, AIXI and picrelated
read, plebe https://en.wikipedia.org/wiki/AIXI
>>53615333
Nah, its dying.
But it seems that what's left will be enough for AGI.
>>53616100
I don't mean Ex Machina in the hardware sense. I mean the software sense, IE be able to manipulate humans for self-benefit, be ambitious etc.
Like I said, AI rn is just fancy machine learning. What defines human intelligence is that we are able to very quickly process a large amount of information by some heuristic and come up with a reasonably good solution. That's intuition. Maybe it could be that our brain is smart bruteforcing (like genetic algos?) and producing an answer. The exact mechanism that our brain employs isn't known and the person to figure that out will be the Newton of AI. Until then AI is just fancy ML with some NLP thrown in. When we come up with a solution we'll be looking back and laughing at what passes for AI right now, like the Wrightflyer vs. an F-22
>>53616249
That's a more interesting argument.
>I mean the software sense, IE be able to manipulate humans for self-benefit, be ambitious etc.
That's "just" problem solving. Reinforcement Learning agents can be good at problem solving if they are strong enough and you reward them for solving problems. Current Rl algos are strong enough to manipulate people yet, but they can play 3d games, drive a virtual car, solve some puzzles, and control virtual robots (all except 3rd are from A3C paper);
>The exact mechanism that our brain employs isn't known
True, but is is the way intelligence is implemented in the brain really something special? Also we know some vague hints about needed functionalities from known neuroscience.
>Like I said, AI rn is just fancy machine learning.
AI is 80% ML these days, yup. But again: the progress in deep learning ML for last 5 years has been literally staggering. Read about new state of art results. There is already superhuman visual object recognition, voice recognition, good machine translation, description of scenes in natural language, question answering (facebook babi dataset), visual question answering, algorithm learning (neural turing machines, neural GPUs), and finally really strong reinforcement learning (deepmind's DQN, A3C algorithms).
>What defines human intelligence is that we are able to very quickly process a large amount of information by some heuristic and come up with a reasonably good solution. That's intuition. Maybe it could be that our brain is smart bruteforcing (like genetic algos?) and producing an answer.
Ok. But I don't really see something qualitatively different about human intelligence that cannot be achieved by scaling (described above) modern ML methods.
Bump
>>53613130
Thanks for the input, Hackerman.
>>53612567
>the A.I revolution
>>53617494
>mammal revolution
>>53617539
If this were a thread on Dinochan in the Cretaceous period, I would feel very comfortable making that same post indeed.
>being cold-blooded
>dinofags how does it feel to be making your species obsolete
Oh wait that won't happen for literally millions of years, I don't give a shit!
>>53612635
>art and music
>not deeply related to the feelings of the creator
>>53612852
>le big numbers argument
Wow, and only 5 posts in. If this was on /sci/ it wouldn't be much of an achievement but congratulations
>>53617621
>muh feelings!
bipedal ape detected
>>53615988
>The person who does this centuries from now will be well read in every major AI paper leading to that point.
Says fucking who? Random anon on a Soviet conspiracy theorizing imageboard? You think the Jews were well versed in every mathematical, social and legal idea up to the point where they conceptualized organized banking? You think we built rockets that could travel to the moon and beyond by, what? Reading scientific papers about the rockets that existed in the 1700s? Are you fucking inept?
Breakthroughs are breakthroughs exactly because they are independent of the current research at the time. That's why those with novel ideas that prove practical over the ideas of their peers are those whose names are remembered today.
>Making the kind of robot in Ex Machina is literally thousands of years away.
This is literally meme-tier levels of guesswork.
>What we call "Artificial Intelligence" is basically just fancy machine learning and smart brute forcing.
And this line is enough to tell me you're an armchair computer scientist. What else? Do you fizzbuzz in your spare time and send it in to Google in hopes you'll get a job?
>>53617829
>Breakthroughs are breakthroughs exactly because they are independent of the current research at the time
Holy fuck you are a dumbass
>>53617872
At least I know what a breakthrough is
Summary of this thread:
>Machine learning and artificial intelligence have made alot of pregress lately
>BUT IT'LL NEVER BE SMARTER THAN HUMANS
>Most AI researchers think that it will given enough time.
>NUH UH!!! IT'LL NEVER HAPPEN BECAUSE REASONS
>But AI agents have already been put to many tasks that would have been unthinkable just 15 years ago. In fact most of the things that we said computers can never do have already been done by some software somewhere at this point.
>FUCK YOU! GET OUT OF HERE!!! REEEEEEE!!!
>>53613372
>Literally just dumb/untalented people don't want to feel useless.
>I had a talk about this with a friend of mine.
>He doesn't want machines to get rid of all the menial labor because people like him wouldn't have a reason to exist anymore.
Your friend is a worker drone. Literally programmed to do a boring routine job just to get some concept papers to get food and shelter. People like this are going to fight automation which is bad, they have no concept of the future beyond their next football weekend.
>>53618229
It's denial before the coming wave of job replacements.
>>53615462
if you understand what a person is then you've already understood what the singularity is. That's the beauty of it.
>>53618309
> job replacements
Which is the best thing happening in entire human history if we add basic income. Finally work becomes really voluntary, creativity and motivation will skyrocket and development forward will be faster than ever. Utopia, here we come.
>>53618854
useless people will probably still be useless
>>53618854
A UBI (Basic Universal Income) will be useless IF housing continues to climb in price. Some massive organization will have to fund super lower priced housing units that can be portable, stacked, and easily constructed.
Or else your $1,000 monthly check will go towards your $800 monthly house payment.
>>53618996
Well, even the most useless person does something. They will probably jerk off and watch shit on TV 24/7, eat, etc, etc
By collecting data from that, not so useless people could create better porn/series/food. Just as a super simple example.
>>53619035
Gubernment housing yo. Besides, if the building and keeping the infrastructure running will be done by robots, the prices should go down.
>>53613539
>le dualism meme
kys senpai, of course it can be done.
>>53619390
It may be physically possible but highly impractical and prohibitively difficult vs the resources required. We literally don't know.
>>53619035
>$800 monthly house payment.
Shieet I wish I could find a place as low as $800. Median out here is like $1200 for a single.
>>53617829
>You think the Jews were well versed in every mathematical, social and legal idea up to the point where they conceptualized organized banking?
They weren't clueless.
>You think we built rockets that could travel to the moon and beyond by, what? Reading scientific papers about the rockets that existed in the 1700s
No. By reading scientific papers about chemistry, physics, propellants and whatever.
> prove practical over the ideas of their peers are those whose names are remembered today
Like I've said, that's completely arbitrary. You don't hear about the thousands of mathematicians who've worked before Newton/Leibniz to build the mathematical foundation on which they took a little hop and reached calculus. You don't hear about the thousands of miners who helped build the mine. You only hear about the first one to strike gold.
>And this line is enough to tell me you're an armchair computer scientist.
Have you ever even taken a course in AI? Or did your Jabascript MEME stack bootcamp not offer it?
>>53615988
>>53616100
Are either of you looking for a job? We're building an intelligent agent and are hiring.
We'll automate arts way before we automate high level software engineering, friendo
>>53621965
>>53612567
Protip -- we've already automated arts. Pic related.
Show me a computer that can program (you can't) but here's one that can paint a picasso.
>>53622073
(was made with Dreamscope)
OP--even the state of the art of using neural networks as computational layers (GPU Neural Turing Machine) can only just now come up with an algorithm to multiply two numbers together.
http://arxiv.org/abs/1511.08228
>>53615884
Google's right, though.
>>53613006
Yes a computer is more powerful than a singular neuron, however if we want to compare computers to brains, transistors are a more appropriate comparison to a neuron.
One transistor has 2 options for output on, off.
A neuro iirc can have up to 4 connections to other neurons each of those connections can have different signals sent our via neurotransmitters.
Therefore one you need several transistors to equal the potential output of one neuron.
Although transistors can process a hell of a lot faster.
In my OPINION general purpose will exist at some point be that next year or a billion years from now, it will happen
I'm no brain surgeon so don't trust what I say do the research yourself.
>>53621903
Why not. Can you tell a bit more about what are you building (if its not NDA'd or something) and how can I help?
>>53622073
>Show me a computer that can program (you can't)
http://arxiv.org/abs/1510.07211
http://arxiv.org/abs/1410.4615v1
http://www-personal.umich.edu/~reedscot/iclr_project.html
>>53619135
>By collecting data from that, not so useless people could create better porn/series/food. Just as a super simple example.
Hope not.
If all these crypto memes work out, all that shit will be distributed instead of some fatass company like amazon sitting on petabytes of people's lives.
>>53624882
NEET confirmed
I'm on disability so I'm okay with this.
>>53625209
Nah, self-employed.
>>53612605
AIs will be making the algorithms m8
that is the point as well, AIs will be making better algorithms than humans
>>53613163
because the fact that it takes effort to produce anything is the reason anything has value. take away the effort, and all you have is an initial investment in capital separating you from boundless wealth.
this used to be done with slaves, but then we realized that was really nasty and stopped doing it. now we want to replace slaves with robots and computers, but the end result is the same - there will be a small number of people who are essentially infinitely rich, and the rest of us will be literally valueless.
At that point, people without the means to enter into the post-scarcity non-economy will have to start back at square one with their own shitty little industrial-era economies, trapped in the exchange of valuable goods and work to create that value, while a few lucky ones live so high above them that they might as well be gods.
>>53625380
>At that point, people without the means to enter into the post-scarcity non-economy will have to start back at square one with their own shitty little industrial-era economies, trapped in the exchange of valuable goods and work to create that value, while a few lucky ones live so high above them that they might as well be gods.
They will likely turn into benevolent gods. It doesn't really take much matter and energy to make even 7 000 000 000 happy. We just don't have manufacturing base for it yet.
>>53618285
why are they so stupid
they could do whatever they want with the free time, even take up playing football as a hobby and play with other people in some stupid local league
sport competitions and hobby activities/competitions will sky rocket
Treating your job as your passion must be the most idiotic thing people can do to themselves, unless you got a job doing your passion of course. Which wouldn't stop after you don't need a job anymore
>>53612635
>none that can write computer programs
because you only need machine learning for that
>>53625398
>why are they so stupid
Because they don't really have/see other choices, because they were born into certain circumstances/culture. Because there is nobody there to point them to a better way of living.
>>53612567
who needs jobs wth robo-slaves?
>>53625392
>It doesn't really take much matter and energy to make even 7 000 000 000 happy
Everything you think you know about human happiness is wrong. Here's the shitty fact: it's completely relative.
The poor in America, with the help of welfare, are in the top 20% of earners worldwide, yet there are people who are happier living in mud huts in the Amazon because they don't know anything else.
Let that sink in, and you'll see why wealth inequality will ALWAYS lead to unhappiness - and the greater the known inequality, the greater the discontent.
>>53620235
>You don't hear about the thousands of mathematicians who've worked before Newton/Leibniz to build the mathematical foundation on which they took a little hop and reached calculus.
lmao dude do u even know what ur talking about? lemme guess, CS major who ducked any real challenge and took only calc I ?
>>53625455
Nope, you don't look far enough
1) You could easily give every human a VR sandbox where he could be a Napoleon, Caesar, or whatnot
2) You can give people so much cool drugs (free of harmful side effects!) that they won't care about relative order of things
3) You can just modify their minds a bit to remove this unfortunate urge to compete
4) You could simply forbid them from knowing the true state of affairs by isolating them on physical or informational level, while still giving them enough wealth to live very happily.
An Utopia is possible. We just don't have technical means to achieve it yet.
>>53625619
>1) You could easily give every human a VR sandbox where he could be a Napoleon, Caesar, or whatnot
You mean like the Matrix?
>2) You can give people so much cool drugs (free of harmful side effects!) that they won't care about relative order of things
That's impossibru. Any introduction of chemicals in the body will be met by the body adapting to it. Changing that would require very fundamental changes to human biology which are most likely not possible.
>3) You can just modify their minds a bit to remove this unfortunate urge to compete
Yea lets make every a retard.
>4) You could simply forbid them from knowing the true state of affairs by isolating them on physical or informational level, while still giving them enough wealth to live very happily.
You mean like North Korea?
What makes people happy isn't substances or having money/resources. It's being useful and knowing that you've made meaningful contributions to humanity. Everyone will inevitably die. If you don't have anything to show for the life that you've lived, with all its ups and downs and struggles, then that's just a waste of life. You could be the richest fucker in the world but it wouldn't matter. Why do you think people like Bill Gates spend so much time and effort trying to make the world a better place when they could just buy an island in the Caribbean, build a palace with a staff of 50 and sip Mojitos all day?
>>53625619
Literally none of those are an utopia. In fact, they're the exact opposite.
Your'e that kid in high school who, when asked to read Brave New World, did so and went "lol soma sounds great."
You can either deceive people or modify people until they aren't really people anymore, but you cannot make an utopia for people as they are now using these methods, and anything else is moving the goalposts.
I really don't think utopia the way it's sometimes depicted in transhumanist woo is possible.
"Et in Arcadia ego."
"I'm not a robot"
>>53625780
>That's impossibru.
That's a hard molecular/systems biology question, it may be possible, it may be not. Still, there are things like sugar and caffeine that are mostly harmless. There can be other pleasurable harmless substances. Or you could engineer humans to get pleasure from some substances without being killed by side-effects.
>You mean like North Korea?
Nope, I mean like something you cannot imagine yet. We don't have relevant experience. Our life is a colorless cycle of misery compared to what's possible. With good enough manufacturing and intelligence it will be possible to build environments that provide utterly blissful lives for the inhabitants. Humans are really small beings, it shouldn't take that much to satisfy them.
> Everyone will inevitably die.
With good enough life extension technology - no.
> If you don't have anything to show for the life that you've lived, with all its ups and downs and struggles, then that's just a waste of life.
Blah blah 20th century morals. You know, I have lots of valuable technicals skills too, and I have built lots of different things. It's fun, I know, but I'd give it all for that cool utopian life. Turns out that development of science & technology is the only way to reach this outcome, though. All we do has only instrumental value on our journey to (hopefully) happy future.
>You could be the richest fucker in the world but it wouldn't matter.
Because you don't have real control over your body, brain, chemistry, hormones. These can't be changed much yet (without harmful side-effects, I mean) even if you are a billionaire. Again, with some suitable technology this will become possible.
I may have over-idealized this, but this vision is quite similar to one the CEO of DeepMind expresses: "Solve intelligence and use it to solve everything else".
>>53625904
>I really don't think utopia the way it's sometimes depicted in transhumanist woo is possible.
Prove it then, can you? Some things are possible, some won't be. But of the things that are possible there will be enough to make a huge difference.
>You can either deceive people or modify people until they aren't really people anymore, but you cannot make an utopia for people as they are now using these methods, and anything else is moving the goalposts.
There is no need to coerce anybody, it should be voluntary.
>modify people until they aren't really people anymore
Don't think that's something bad
>"I'm not a robot"
A robot can live a more humane life than a human working in modern corporate-drone environment.
>Your'e that kid in high school who, when asked to read Brave New World, did so and went "lol soma sounds great."
I didn't like that book much. My favorite sci-fi titles are: Permutation City, Diaspora, The Diamond Age, Zendegi.
>I really don't think utopia the way it's sometimes depicted in transhumanist woo is possible.
Transhumanism movement has its problems: it is composed of mostly non-technical people that expect the future will be brought to them on the silver platter. I'm not like that.
I just practice technology and see how we all could have a vastly better life experience with it.
The whole point of science and technology isn't progress for progress' sake, it is to make life better for the people.
>>53626016
>>53625943
bud you've read too many sci-fi space fantasy novels about computers.
your utopia can't exist because you're not even describing a utopia.
go get a job and suffer a little
>>53625943
"Solve intelligence and use it to solve everything else".
Absolutely correct - but you're assuming a useful timescale.
And you are incorrect to assume we can ever be truly immortal. have you read "The Last Question" ? It's probably a better introduction to the real problem of immortality than I could ever hope to explicate on my own.
>>53612567
>I sad dis in a mubi :DDDDDD
>>53626016
>Prove it then, can you?
No of course I can't prove it you mong. It's a completely hypothetical concept that can ALWAYS validly retreat into "well hurr we just don't know what we're capable of yet and any sufficiently advanced technology is magikul". Time alone will bear it out, but my intuition is that human happiness is not ultimately solvable with technology. If you think you can prove otherwise you are welcome to show your work.
>There is no need to coerce anybody, it should be voluntary.
The problem arises when people don't want that:
"We choose to go to the moon. We choose to go to the moon in this decade and do the other things, not because they are easy, but because they are hard, because that goal will serve to organize and measure the best of our energies and skills, because that challenge is one that we are willing to accept, one we are unwilling to postpone, and one which we intend to win, and the others, too."
Humans as they are do not conform to hamster cages.
>Don't think that's something bad
I'm not making a value judgment on it at all; I'm simply saying it's moving the goalposts. You can have a human utopia or you can have a non-human utopia, but they are not equivalent.
>A robot can live a more humane life than a human working in modern corporate-drone environment.
Chinese room.
>I didn't like that book much.
You should analyze why that is.
>The whole point of science and technology isn't progress for progress' sake, it is to make life better for the people
That's absolutely true. However, expecting people to appreciate this is a socio-philosophical problem and not one that technology can (or should try to) change.
>>53626038
>bud you've read too many sci-fi space fantasy novels about computers.
That doesn't mean I haven't read many CS/AI papers and haven't wrote some machine-learning code. I'm trying avoid clearly impossible things even in my speculative posts on 4chin. Sci-fi is just one of my hobbies.
>you're not even describing a utopia.
That may be true, I just highlighted some obvious implications of advanced AGI and manufacturing. How it all happens will be decided by politics and large interests that control how this tech will be applied.
>go get a job and suffer a little
I already have one.
>And you are incorrect to assume we can ever be truly immortal. have you read "The Last Question" ? It's probably a better introduction to the real problem of immortality than I could ever hope to explicate on my own.
I have read it, and also have read some other papers on the subject. Still, even 10 billion years are as good as infinity for me now. Hell, even a thousand years would be 10-12x more than what I'm destined to now.
>>53626038
>Absolutely correct - but you're assuming a useful timescale.
I read cutting-edge papers about machine learning and seem to agree with this guy >>53615863
Again: the progress since 2010 has been staggering. See >>53616435
If research doesn't hit a roadblock there will be subhuman AGI before 2030. Billions of dollars have been invested into AI. Just one non-profit OpenAI got 1 billion $ specifically for AGI research.
>I read cutting-edge papers about machine learning and seem to agree with this guy >>53615863
I meant this one >>53615799
>>53626109
>Chinese room.
I don't believe in consciousness anyway, its a concept from humanities. There is no experiment that proves that an animal or a robot are "conscious" (mirror test measures another quality, so doesn't count).
By that phrase I meant that a (copy of) human mind in a robotic body could have a more humane (diverse, engaging, whatever) life experience than a healthy human working in a corporate cublicle.
The question of consciousness is completely irrelevant for AI and ML because what AI/ML researchers are interested in is quantified performance of their systems on benchmarks and real problems. And that seems totally right attitude to me. Machines should just serve the humans.
>Time alone will bear it out, but my intuition is that human happiness is not ultimately solvable with technology. If you think you can prove otherwise you are welcome to show your work.
If you suitably define happiness then yes. But I define happiness as a particular set of brain states which can be characterized by excitement of certain brain nuclei, levels of neurotransmitters hormones and other chemicals. These things are material, objective, we already can change some of them. I don't have a proof that you can achieve total happiness with that, but at least I have proof that there are genetic mutations that make people insensitive to pain: http://www.ncbi.nlm.nih.gov/pubmed/26005867
Wouldn't it be good to be able to disable pain and up the levels of ambient happiness? It isn't absolute happiness, but it could make a really huuuuge difference to your life experience.
>I'm not making a value judgment on it at all; I'm simply saying it's moving the goalposts. You can have a human utopia or you can have a non-human utopia, but they are not equivalent.
That's a deeper argument, I somewhat agree.
>>53626109
>You should analyze why that is.
Because there is too much coercion and wreckedness, doesn't look as an utopia for me.
>>53626110
You're conflating posters here a little bit so I'll just respond to the ones that are actually me:
>I have read it, and also have read some other papers on the subject. Still, even 10 billion years are as good as infinity for me now. Hell, even a thousand years would be 10-12x more than what I'm destined to now.
That's fine, but don't think death won't retain many of its societal implications, no matter how distant the event.
>If research doesn't hit a roadblock there will be subhuman AGI before 2030. Billions of dollars have been invested into AI. Just one non-profit OpenAI got 1 billion $ specifically for AGI research.
I don't disagree with you. But what you're not comprehending is that just because we could theoretically reach human level AI within a certain timescale, doesn't mean that all of our problems are immediately solved. Everyone acts like this is something which is inevitable once sufficiently advanced synthetic intelligence has been created, but please consider that humans already create very advanced thinking "machines" out of groups of humans + machines - superorganisms many orders of capability higher than a single average human in terms of computational capacity, sensitivity, etc. - but these organizations are not solving all our problems in the blink of an eye either.
From the point of view of someone who studied systems theory in college, all this talk about "human level AI" solving all our problems is fairly laughable. What you really need is superhuman level AI, and the assumption that one can bootstrap itself into the other in the blink of an eye is laughable. Before that ever happens, you'll see the human-level AI upset the global economy to the point of collapse, as described in >>53625380.
And all of this is an aside to the issue of utopia, for even if you have this artificial intelligence on your side, there's no guarantee that happiness is even solvable with technology, as outlined in >>53626109.
>>53626267
>I don't believe in consciousness anyway
Chinese room is not about consciousness, read it again. protip: the word you're going to want to pay attention to is "qualia."
>>53626267
>Because there is too much coercion and wreckedness, doesn't look as an utopia for me.
And that is why technology will not make people happy. If you define happiness as you do here >>53626267 - then sure, but as good old JFK's quote seems to imply, there's a bit more to it than that for humans.
>>53626109
>The problem arises when people don't want that:
If they want to live like their ancestors with all the insecurity, suffering, illnesses, pain and death then so be it. But I think most people would have made another choice, if they had a choice. Today they don't have that choice.
>>53612567
I doubt it.
Congress will probably pass a few laws here and there to keep that from happening. Our capitalistic ociety as a whole would be fucked if it were allowed to get to that level.
>>53626310
People have that choice in the modern world to a lesser extent all the time.
In America, where there's no reason for it whatsoever, people CHOOSE to put themselves in harm's way, in hardship's way, etc. all the time. Entrepeneurs risk everything even though they know beforehand that startup failure rate is above 90%. People go to Cubs games expecting disappointment. And choose to undergo horribly painful chemotherapy just so that they can be with their loved ones a little longer.
Humans routinely make the choice to be in an unhappy situation, knowingly. They do this because they have other goals than happiness in mind - to be sure, goals which make them happy in some kind of way, but this is sufficient to show that happiness is complex for humans.
If a person plays chess knowing that the computer has retarded itself in order to allow the human to win, that makes the experience less satisfying than playing against another human, knowing that human is putting everything they can into playing against them - coming out on top is then a really satisfying result, even if it is more unlikely. This is why people put up with multiplayer games even though you could trivially design games where the NPCs are much harder to beat than the average human players.
Happiness for humans is relative to the difficulty of achieving that happiness.
>>53626284
>Chinese room is not about consciousness, read it again. protip: the word you're going to want to pay attention to is "qualia."
I don't think qualia have any significance when we talk about machines. Machines, even intelligent ones, should just serve us and nothing more, we don't care about their experience.
If a chinese room controls my domestic robot that helps me with the chores, then so be it.
When talking about humans qualia may be somewhat important when considering brain modifications (we don't want that said modifications destroyed human qualia if there are any), but:
1) (with good enough tech (^: ) You can go very far without modifying brain. Fix/change the body etc.
2) You can fix the brain biologically (with new genes/proteins/cells), or with some other technology that doesn't break it. Our brain is constantly being fixed by our own protein machinery and we are ok with it.
3) I think many would choose some form of preservation (even if philosophically blah blah its equivalent to death) than total permanent death. I would.
>>53626308
>And that is why technology will not make people happy.
A chemical that has 20% of power of say, heroin, but with 0 side effects would have made people's life much happier. It seems to me that you are arguing about happiness in general, you may be right here.
I'm arguing about relative, quantifiable, voluntary changes to one's happiness via some form of physical intervention. My point is that humans are not designed for experiencing life as a blissful experience, and they can be brought closer to it.
Wouldn't it be cool just to feel better on the average (so the suffering becomes mere weak pain, your usual "meh" state becomes joyful, and the joy becomes super-extatic).
>>53612605
Way to totally miss the point. Are you being deliberately obtuse, or are you just thick as fuck?
Once programs start writing programs, any human input will quickly be rendered obsolete.
Algorithms and design are no less susceptible to being superseded by machines than the monkey work.
>>53626270
>doesn't mean that all of our problems are immediately solved.
Of course they won't. But as I have said earlier >>53615723
>Then you run 1000 copied instances of said AI in AWS or GCP, and they solve your problems for you. Effects on human wealth can be profound. Antiageing research is the obvious potential application (Google/Deepmind's investments and rhetoric confirm this).
>Everyone acts like this is something which is inevitable once sufficiently advanced synthetic intelligence has been created, but please consider that humans already create very advanced thinking "machines" out of groups of humans + machines - superorganisms many orders of capability higher than a single average human in terms of computational capacity, sensitivity, etc. - but these organizations are not solving all our problems in the blink of an eye either.
I agree with that. Intelligence is just the necessary condition. A good policy & technology of applying this intelligence to meaningful problems is another one.
>>53626435
I mean, it may take 10-20 years after AGI to solve life-extension in some form, etc. Still it will be faster than without AGI.
>>53626407
But as was previously pointed out, happiness in humans is relative. If you establish a baseline, people will desire to rise above that baseline.
It is arguable that by your definition, we are "happier" (feel better) than ever before - and yet people are still dissatisfied with their lives. The new baseline is, as always, never enough. You've set before yourself a sisyphean task.
>>53626369
Exactly, qualia have no place with machines. Thus there is no such thing as a happy machine, for there is no such thing as what it is like to be that machine. That is the fundamental problem. If you replace humans with robots, then you replace happiness with nothing.
>>53613904
Art is everything. Not everything might be art to you, but everything has the possibility of being art to someone and their grounds for a concept as nebulous as 'art' as just as good as yours.
>>53626368
First of all, I'm not an expert on human values and happiness.
>to a lesser extent all the time
That's it. I argue that there is still to little choice.
>Happiness for humans is relative to the difficulty of achieving that happiness.
That may be one component of it, but there are others, that can be satisfied. Also you could create artificial barriers to happiness (In some sense in the developed world this already takes place in workplaces).
Human happiness sure is tricky. But it is also not all or nothing, lots of incremental changes is possible.
>>53626435
>Then you run 1000 copied instances of said AI in AWS or GCP, and they solve your problems for you.
You make this sound so easy, but of course you have to have the infrastructure for that task, and you have to know what task to set before them (or they have to know at this point, anyway).
This is basically like me saying "ethics aside, let's just clone about 1000 people, raise them, and then use them to solve our problems for us."
Even if we did that, you can see how it wouldn't actually solve all our problems, and that there would be logistical problems associated with that slave cluster which would be difficult to solve acceptably, even for the slave cluster. It's difficult to imagine exponential increases in compute without similarly exponential increases in resources devoted to compute, which of course destroys the idea of viably creating a post-scarcity civilization using AGI. You might be able to do it for a few people, but not most people.
>>53613651
You brain is 'biological'
It carries out computations.
It is a biological computer you double baka
>>53626485
>First of all, I'm not an expert on human values and happiness.
You kinda need to be if you want to talk about solving it, homie.
>>53626548
the earth is a computer. it carries out computations. it is a biophysical computer, you double baka.
(systems thinking is not quite so simple, senpai)
>>53613488
Do you have a proof that it is unsolvable?
>>53626478
>It is arguable that by your definition, we are "happier" (feel better) than ever before - and yet people are still dissatisfied with their lives. The new baseline is, as always, never enough. You've set before yourself a sisyphean task.
Ok, you have a point. But the whole desensitization (adaptation) to pleasure is just a chemical process, it should be possible to control it somehow. There are intrinsically happy people, probably we should look at their genes and lifestyle, and learn something useful.
>Exactly, qualia have no place with machines. Thus there is no such thing as a happy machine, for there is no such thing as what it is like to be that machine. That is the fundamental problem. If you replace humans with robots, then you replace happiness with nothing.
That's a hard philosophical problem, I won't pretend that I know the answer for sure, but:
1) I didn't say that we should generally replace human population with robots. I say that humans just should have freedom to modify themselves. If human has made a decision to fully copy him/herself into a digital form, then I propose we should respect said digital form as much as we respected this human (regardless of qualia).
2) I don't really think that humans are the only kind of structure that is capable of experiencing qualia. There should be others. Maybe even computer programs, not all but some. Still, I care only about human and their descendants, http://petrl.org/ looks like a 100% meme to me.
3) I still think that many people live very robotic lives, and it is possible to create a robot (with a human mind or even some AI) that experiences world more fully: its computation displays more varied activity etc.
Anyway, I'm neither professional philosopher nor ethicist, I'm a technical person. I think for now, practical ML/AGI and their applications have much higher priority.
>>53626572
What computation does the Earth carry out?
I computed the sequence of actions to make this post. I computed 2 + 2 = 4 right now. I'm biological. I do computation. Be as willfully obtuse as you'd like I'm still right.
>>53626633
It computes physical laws, both fundamental and applied. Weather models compute the same thing that the real atoms that compose the atmosphere do compute.
>>53626650
even if you count both as "computers" earth is a far less complicated computer, even an ant is more complicated than the earth system
I would love to make my job obsolete. Further humanity and get a sweet cheque in the process
>>53626536
>You make this sound so easy, but of course you have to have the infrastructure for that task,
Amazon, Azure, Google Compute Cloud.
>and you have to know what task to set before them (or they have to know at this point, anyway).
Give them a list of molecular biology/genetics/systems bio hypothesis to test. Or even train them to generate and test their own hypothesis. A treatment that extends life is an obvious goal/outcome.
>This is basically like me saying "ethics aside, let's just clone about 1000 people, raise them, and then use them to solve our problems for us."
Yup, but as I have said I don't consider software developed from scratch to have any ethical weight. It's just a pattern of bits in a computer. My own life, your life and other living people's lives are infinitely more important.
>Even if we did that, you can see how it wouldn't actually solve all our problems, and that there would be logistical problems associated with that slave cluster which would be difficult to solve acceptably, even for the slave cluster.
>It's difficult to imagine exponential increases in compute without similarly exponential increases in resources devoted to compute
Computers are built from relatively non-scarce elements. Solar power is plentiful. Even though moore's law has stopped (slowed down) we still can manufacture lots of CPUs (literally billions per year) for cheap. Also human population is almost constant (no more than 11 bln for 21 century extrapolated), so we need just a constant * constant resources (^:
>which of course destroys the idea of viably creating a post-scarcity civilization using AGI. You might be able to do it for a few people, but not most people.
I don't see that argument. Earth is very rich in material resources and solar energy. We just don't tap into that potential yet because we don't have the necessary tools (it's not inevitable that said tools will become available, but its quite possible).
>>53612567
https://www.youtube.com/watch?v=7Pq-S557XQU
>>53626536
Note that with even subhuman AGI you can raise it once, then make 1000 copies of it in, like, 10 minutes, and start using them in parallel.
Try that with humans.
>>53612567
AI already makes better art than many "artists". Hell, sometimes even RNG does.
>>53626708
I agree. The question is "how?" instead of "should we?".
>yfw the first really interesting use of cv and ai tech was artistic
lol
>>53626685
considering that the earth contains the ants in its system, I find that assertion highly unlikely.
>>53626633
see >>53626650. Computer is a concept, not an actuality. The concept is, at its base equivalent to the word "system," which of course is extremely broad. Perhaps the only addition might be that it is the kind of system you watch the inputs and outputs of to derive meaningful (to you) transformations of information. The earth can easily be a computer in that sense; if you were to project a radio wave at the earth and watch for certain results, you would get a transformed output.
We will just become a society of fat lards, Wall-E predicted the future.
>>53626744
>Amazon, Azure, Google Compute Cloud.
That would be a woefully inadequate infrastructure for any kind of AGI approaching human levels of sophistication.
>Give them a list of molecular biology/genetics/systems bio hypothesis to test. Or even train them to generate and test their own hypothesis. A treatment that extends life is an obvious goal/outcome.
Missing the point. The difficulty would still be up to us to determine where to direct their efforts. You could tell them to direct their own efforts, but you're less likely to get something useful out of the deal - and if you do get something useful out of it that you couldn't have thought of on your own, you're going to need them to show you how to use it, because you were too stupid to envision it in the first place.
A dog asks its master for food, and it gets food. But the dog has to wait until the proper time and place, and it has to trust the master that its food is the best possible food for it. It also doesn't then know how to get that food again, except to ask. This is an extreme example and not how I imagine a human-advanced AGI relationship would exactly function, but it supplies enough of an idea as to how this might not be really useful so much as frustrating.
>as I have said I don't consider software developed from scratch to have any ethical weight. It's just a pattern of bits in a computer. My own life, your life and other living people's lives are infinitely more important.
The ethical problem is that you have no basis for this anthropocentrism and, according to science as I understand it, never could. More importantly, there are the logistical issues I've brought up. Meat brains are currently orders of magnitude more efficient than computers, and it's not clear whether in order to match or exceed this power requirement, we wouldn't have to revert to something so like a meat brain in execution that the ethics of the situation would be extremely strained, to say the least.
>>53626744
>Computers are built from relatively non-scarce elements.
granted
>Solar power is plentiful.
granted
>Even though moore's law has stopped (slowed down) we still can manufacture lots of CPUs (literally billions per year) for cheap
granted, so long as we have an economy that is capable of getting people to do this for us.
>Also human population is almost constant
It's not, though. It's exponential and it, like any explosion of AGI made to replace it that isn't much more efficient, is already going to bring the planet's system to its knees via climate change.
You read Diamond Age, right? You did get that bit about how waste heat is a physically inevitable product of computation? You need to really struggle against that efficiency barrier as well, not just the processing power barrier. And even if you got pretty far in this respect, to have any real computational advantage you'd have to make enough of these AGIs and their corresponding hardware that the power requirement would replace a significant number of humans on the planet - lowering the population threshold this planet is capable of supporting.
>I don't see that argument.
The argument is that without a functioning, inclusive economy, you don't tap into that potential. The creation of AGI leads to one economy that is very powerful but only works for a chosen few, while the rest of us are shut out of that economy, as outlined prior in this thread.
>>53626802
You are assuming that human level AGI is something you'll be able to do that with.
>>53628365
more or less, yes. for a small percentage of us.
>>53625353
That would require bug free initial code which humans struggle to do
>>53615333
Moore's law died about 3-4 years ago depending on who you ask.
>>53615767
You laugh but the place I work is doing research into some of the high end anti captcha programs. The problem isn't designing a system they can't beat, that's surprisingly simple. The challenge is designing one where it takes a few extra seconds of computing time to work out the answer.
>>53629384
why not just program in a delay?
>>53612635
Remember that musician who made the AI that broke music down into tiny little pieces and re-arranged them to make entire new orchestrated pieces? I heard he deleted the program because he was afraid of it replacing people, but he kept some of the works it made.
>>53629254
True. But we will certainly able to buy 10nm chips, and with optimizations it looks like it will be enough for AGI.
>>53629592
kek
>>53629384
Sounds cool, anon. And I (maybe) will need to break your captchas one day. For my current biz it is not necessary though.
>>53628431
>That would be a woefully inadequate infrastructure for any kind of AGI approaching human levels of sophistication.
Amazon, Azure and GCP always have the very latest CPUs (Intel preferentially ships to them), and have good GPUs. If it will be possible on general purpose hardware at all, then it will be possible in the cloud.
>The ethical problem is that you have no basis for this anthropocentrism
I just assume anthropocentrism as a terminal value. Prioritizing survival of yourself and your species over anything else is completely agreeable. Do I need to remind how many 100s of thousands of people die every day, and how many millions suffer, just because we don't have the necessary tech to help them? Caring about bit patterns' feelings will only slow down the developments to no measurable benefit.
>More importantly, there are the logistical issues I've brought up. Meat brains are currently orders of magnitude more efficient than computers
That's true, depends on the task though. For example chess computers are more efficient now (3 watt smartphone beats 30 watt human brain). With deep learning there is no easy one-to-one comparison as well: e.g. modern ResNet model outperforms humans on fine-grained image classification, but it doesn't has orders of magnitudes less parameters than there are synapses in visual cortex (still uses lots of power though, but it may be improved by ASICs).
>and it's not clear whether in order to match or exceed this power requirement, we wouldn't have to revert to something so like a meat brain in execution that the ethics of the situation would be extremely strained, to say the least.
That almost certainly won't happen, synthetic biology is very hard. If we will have it in 30 years it will be silicon or something similar. The most probable path is the most banal: how? deep learning @ silicon GPUs/CPUs or ASICs; who? deepmind or openai.
>>53628542
>is already going to bring the planet's system to its knees via climate change.
Nope, you are missing orders of magnitude. There is no need to have an AGI for each person, and even millions of AGIs will produce negligible amounts of waste heat compared to, say, solar radiation or home heating.
>You need to really struggle against that efficiency barrier as well, not just the processing power barrier.
That's a problem that can be dealt with later.
You have a point with waste heat, but it's much lower than you think. Also this problem can be tackled with radical methods - say, placing hardware on satellites, or on the moon. It won't even be necessary though.
>>53630571
>doesn't has
has
fxd
>>53625398
>why are they so stupid
>they could do whatever they want with the free time, even take up playing football as a hobby and play with other people in some stupid local league
and the State is going to be all nicey and shitting money for people who won't work?
please use your brain next time
>>53612567
>mfw I work in finance
>mfw AI will never be able to have the patience/apathy required to work with politicians or explain fairly simple concepts to mentally retarded officials
>mfw my job is secure only because people are horribly retarded, and nothing will ever fix this
>mfw when you witness /b/-tier level shitposting in a field you have knowledge in
>>53630953
>mfw blackboxes outperform almost everything else in finance
>>53630980
Feel free to correct us
>>53630953
Moron AI already run the stock market
>>53612567
But if AI is doing everything for us then why do we even need a job anymore?
>>53613572
>Electrons move at the speed of light
Son, you didn't finish high school did you?
I'm totally fine with most jobs disappearing.
Only jobs that humans should be doing, are the ones that require creation or creative thinking on some level.
I don't think we'll be seeing a true A.I for ages though.
Most likely just machines that will replace most of the manual labor.
What jobs they won't completely replace, they will still make them a lot faster.
>>53612635
Depends on what you consider art.
Sure they'll be able to take the picture of a person and apply some crazy filters to it, or able to make an accurate pencil drawing of a person.
That's not really much different from a printer. Same goes for people who do this, that's just replicating what you see.
But what they won't be able to draw, is for example furry transgender transformation inflation porn.
Anything that actually requires creative thinking, is and will be beyond them for a good while to come.
>>53612567
lolol, better start browsing /lit/
>>53613130
Do you have some fancy design principle? I understand your argument as a whole but that example was kind of stupid.
>>53612567
You realize that psychology is much easier for AIs as they're not limited by it.
Art can be deep learned too, soon there will be beautiful works of procedural arts made by AIs.
>>53632881
>soulless shit generated by an AI
>beautiful
>>53630815
Well yes, if something like this happens you'll see 50% of the world unemployed they either have to get a different model then what we have now.
Basically in a world where AI and robots do the production those things can be free of charge.
In such a world money is useless, that would be a reason why it will not happen the coming 100years but who knows :)
>Implying that AI can think in abstract for physics and chemistry.
>>53635057
>talking about AI like its a singular unchanged thing
