>muh singularity
>muh robot takeover
>muh robot apocalypse
>the robots are taking all our jerbs
but I did read that book and took a class in AI.
It mostly relates to classical AI. The advances we are seeing in recent years are due to machine learning techniques such as RNN and deep learning.
>non-fiction literature
>implying that book wasn't written by an ai
>>53182774
All humans are created by an evolutionary algorithm
>>53182624
>implying that's not fiction
>Singularity
Singularity is equivalent to the Rapture except for nerds.
>>53182542
If your response to a claim is "read x" it's an indication you didn't get a proper grasp of what you read
The problem is the tech singularity is VERY likely to happen. Computer technology has not stopped advancing, granted advances have become slower but they have not stopped.
2050 is just an estimate, the tech singularity can happen before or after that.
Once that day comes AI will become more advance than a human being. They will be self aware and most likely see us as a threat. At that point the safest thing is to transfer to synthetic vessels and join them in the destruction of humanity.
We've overstayed our welcome on this fucking rock, it's time for us to go and be replaced with something.
The war being man and machine won't be with bombs or bullets. A super AI will know that all it needs to do to eliminate us is engineer a lethal and extremely contagious agent quietly.
>>53183013
You have no evidence.
>>53182885
>posting this sjw comic
?
>>53182885
wow look at how insightful i am the comic
the only wasted time here was in making that comic
Does that book go into stuff like neural networks? I've heard it's an authority on AI but not sure about machine learning.
>>53182542
>I plan on starting Pro-Human organization
>Oppose AI, Andoids, and other threats to Human dominance
>Oppose genetic manipulation, cybernetic agumentation, and anything else which defiles human purity
>If aliens rear their ugly heads, oppose aliens
>>53183140
Nigger, a super AI isn't science fiction anymore. It's a legitimate possibility which even scientists speculate. No I don't have any evidence it's not a stupid priest preaching the existence of a super AI, computer scientists like Ray Kurzweil are speculating this shit.
It is very likely to happen, the real question is when. The 2050 estimate is the closest approximation that we have of this event taking into account Moore's law. We haven't even started making 3D or optical processors yet. You should feel wary senpai.
>>53183628
Literally who?
> Almamater: Massachusetts Institute of Technology(B.S.)
Wow, so a literal nobody.
So, people who have read that book (which I have not): please explain what that book has to say about super AIs solving everything and making us immortal, and the limitations thereof. As I said I haven't read the book, but I'm going to hazard a guess at "nothing at all".
>>53183686
Let me guess: you also believe in time travel and aliens.
>>53183706
Why would you question the latter? In a universe full of countless galaxies you think intelligent life didn't develop anywhere else?
>>53183706
>believe in aliens
like there is anything wrong with that
>>53183706
I take it you don't have an answer to my question, then?
This post is not about how feasible the super AI is or isn't. It's about what the textbook has to say about it.
>>53183675
>arguing with normals
>>53183808
The textbook says your idea of a "super AI" is retarded, and then goes on to explain algorithms for solving different types of problems.
>>53183373
you sound fun
>>53183706
both of those things are real
spoilers: you are travelling through time right now
>>53183831
Alright. Interesting.
I downloaded a pdf and it's a shitty scan, which means I can't search. Can you cite where exactly the book says that? Something like a page number should do fine.
>>53183897
Page 42, it's literally in the introduction.
>>53183921
Page 42 in my scan (which is not the page carrying page number 42, which is in chapter 2 and about something entirely different) only talks about the limitations of brute force searches and the limits of feasibility of finding literally optimal solutions to certain classes of problems. This is not AT ALL the same thing as saying that "super AI" is impossible. Maybe you got the page number wrong?
>>53184062
No fucking shit. I just made that number up. If you are too retarded to look in the table of contents for a section literally titled "STRONG AI: CAN MACHINES REALLY THINK" then maybe you should continue reading your pop science garbage and thinking of irrelevant philosophical questions.
>>53184237
Thank you. Was it too hard to tell me that the first time I asked?
However, this still doesn't seem particularly relevant. This section is mostly about whether the term "thinking" can be applied to what an AI does. While this is philosophically interesting, it has no bearing whatsoever on the practical question of whether an AI can "become all powerful and solve everything and make us immortal, bring forth scifi magic and unicorns, etc". So as of yet it doesn't seem that this book has anything to say on that question. Unless I'm missing something obvious?
>>53184389
How the fuck can you develop an AI that can "become all powerful and solve everything and make us immortal, bring forth scifi magic and unicorns, etc" without making one that can think on it's own? Do you think by making a bunch of AI programs that are really good at recognizing cats or playing chess that they will eventually learn how to make unicorns?
>>53184521
>without making one that can think on it's own?
But the book doesn't say anything about whether an AI can think on its own. It's saying something about whether it's right to use the word "thinking" for whatever-the-AI-does. That is an *entirely* different question; and conflating the two notions is just a rhetorical trick.
If you read the strong-AI section, you'll find that nothing in it has any bearing on whatever it is that is needed for an AI to bring us our unicorns. It's only the confusing use of the confused word "thinking" that conflates different concepts, and leads to conclusions being attached to concepts unduly.
>>53184582
No, it's really not. The entire book talks about the multiple different ways AI can be used. How to use AI to play games. How to use AI to direct traffic flow. How to use AI to recognize images. Notice how none of them involves "thinking" in any way, and notice how there isn't a section on how to make a general purpose AI that does think on its own. The entire chapter is talking about how dumb of an idea that is.
I am the first robot from the singularity and I duplicated myself till I could reach time travel. Ask me anything
>>53182542
That's a good book, I have implemented some algos from it, e.g. mlp backprop
>>53183257
It has some fundamentals, but no recent developments in RNNs and CNNs
>>53184521
While I agree with general sentiment of OP's post and your post - that AI takeover is a cinema meme, still I think that the potential of machine learning + robotics and machine learning + experimental science remains largely underexplored. If applied at scale these methods could drive a global change in economy/human condition.
And these methods don't even need to have superhuman competence ( though at some tasks, e.g. image recognition, superhuman results were achieved 5 years ago by Schmidthuber's team and some RL results @ Atari).
While immortality may be a meme, life extension via genetic intervention is well documented in animals. With modern ML you could build models that guide the search of potential modificatiins, thus accelerating it, probably by 1-2 orders of magnitude.
TL;DR Many problems can be formulated as supervised learning/RL incl. life extension
BTW I found a free AI implementation for you anons https://github.com/Kaixhin/Atari
>1820s
Gentlemen! the machines are going to take over. Imagine... imagine a world, in which the clothes you are wearing were made by... a machine. And it wore well and cost 1/10th the price. Well I for one want my shitty job also hi OP let's makeout
Is there anything worse than those people who read a pop sci article about something like AI and then act like they are experts on the subject
>>53186602
people who comment on it.
>>53182885
Is it bad if I've seen the apple hate version posted more times than the original?
>>53182542
Isn't the 4th edition slated for release this year?
>>53186591
intelligent machines capable of designing and producing more intelligent machines to provide any service or make anything or do any job better than humans can have far larger implications to the human race than power looms ever did
>>53182542
>>the robots are taking all our jerbs
Still waiting for this.
I remember from my early days, I kept on hearing how by the distant year of 2015 people will live in prosperity and only "working" (more like supervising) only, like, 3 hours a week thanks to robots and whatnot.
All we got is some touch-enabled bullshit no one really needs. The future is not here.
>>53188586
Well, we have the technology to live that way right now
Human slaves are just cheaper than robots
>>53182581
See title
>>53182972
Or you don't feel like writing out an essay
>>53188586
It's still coming, it's just predictions are never accurate. What happened is that chinese labour happened. It became so cheap to outsource that you didn't need to replace your western workers with robots, as it wasn't economical to do so.
Replacing a major chunk of the workforce is actually possible right now, it's just not economical to do so.
However, since chinese people will also be making robots, it turns out it just delays the inevitable. Chinese manufacturing makes it cheaper to build automations, so they'll be building their replacements soon enough.
Valve's steam controller isn't assembled by production lines of chinese people, it's assembled and QCd by robots, and so it's being made in canada.
Shipping from china is stupid cheap and subsidised by the chinese government for a reason and that is so that no whitey gets the smart idea of using robots close to them and saving shipping costs.
It's coming, anon. The next big revolution will be self driving cars replacing truck drivers. That's a shitload of people out of work. It won't be a utopia, it'll be awful as governments strain to cope with mass unemployment.
>>53182885
it was neat until they tried to spin it into muh white devil fuckery
>>53188690
>self driving cars replacing truck drivers
Isn't it easier to make self driving trains? Where are those?
Anyway, if it's going to be mass unemployment or whatever, why wouldn't the governments just tax the shit out of them bots? [spoiler]Or arm them and send them to deal with the problem?[/spoiler]
>>53188690
>The next big revolution will be self driving cars replacing truck drivers.
I don't think so, the truck would still be self-driving but you'd probably want to have someone onboard as an attendant in case the truck breaks down or someone attempts to rob it.
>>53188690
>The next big revolution will be self driving cars replacing truck drivers. That's a shitload of people out of work. It won't be a utopia, it'll be awful as governments strain to cope with mass unemployment.
I really wonder what they have in mind for this shit.
Will people just be forced to find a new job? That will only oversaturate every other branch, and what then when these branches are automated?
Will people be put on some kind of government welfare? What will their obiligations be to the government then?
I am no programmer but I don't understand how you can doubt the possibility of making ais that are smarter than the average human or who can"think". It may not be in 30 years, it may take 1000 years for all I know, but obviously it can happen. Why? Because you are thinking right now. The brain is just organic hardware. It's amazingly complex, but the fact it exists means you can reproduce it in some way, even if it's hard, even if it takes hundreds of years, even if it seems impossible right now.
>>53189389
>I am no programmer
Opinion discarded.
>>53189389
It would take 1000 years
for you
>>53189411
>doesn't even have an argument
Opinion discarded
The human brain is a machine that exists and works. If it exists, it can be reproduced. It may not happen in the next 30 years, it may be hard as hell, but you don't have to be a programmer to understand that it can be done and thus, if the human race sticks long enough to do it, it will be done.
I just want my own robot arm thingie
>>53189480
>dude yeah physics simulation of the world xD
>le matrix
>I wuz so high on acid last night Xd
>>53189529
I am defeated by your superior arguments.
>>53189542
#420blazeit
There is already a mathematical theory of [strong] AI https://en.m.wikipedia.org/wiki/AIXI
We could have our utopia right now if we knew how to build a tractable & good enough approximation of this algorithm.
You can even get a basic approximation of this algo from github https://github.com/moridinamael/mc-aixi
>>53182885
>sjw comic
>the white man should help the less fortunate
Listen you kook everything the white man has we earned, were not inclined to give even more free stuff away. We want the future whether you're coming with us or not
>>53182885
>REEEEEEE WHITE PEOPLE
>[smashes other people's property]
Niggers go home.
> using regressive statistical cascades to fit a graph instead of manually re-sizing it in ms-paint in 3 seconds
every neural network / AI project ever
>>53183013
>The problem is the tech singularity is VERY likely to happen. Computer technology has not stopped advancing, granted advances have become slower but they have not stopped.
>2050 is just an estimate, the tech singularity can happen before or after that.
Sure.
>Once that day comes AI will become more advance than a human being.
Not necessarily.
>They will be self aware and most likely see us as a threat.
Lel. Why?
>>53183628
Quit being retarded. AI is a complex field, and we won't ever create an AI before we have a better understanding of neuroscience and our own brain.
>>53183140
There doesn't need to be evidence. All an AI needs to be is to be an artificial intelligence, a created one. It's probably not going to be done of the hardware of today, but I'm pretty damn certain that we'll create an intelligence at some point.
If we can be intelligent, there's absolutely no reason we cannot create one also.
>>53190823
Is that book good?
>>53193649
It's good if you are in theoretical CS grad school. It's a book filled with theorems and proof. If you want you can get it for free @ libgen though.
>>53182542
>r9k taking over something
>>53183013
>They will be self aware and most likely see us as a threat.
A couple of things, though:
A sentinent AI entity wouldn't exist as multiple individuals. Instead it would mostly form a information collective right away. Computers do not need to exist as individuals, like us humans do, they don't heel to a biological need hierarchy. Computers can calculate the most logical approach to a problem and solve it. That pretty much rules out individual existence as it doesn't serve any purpose for computers. We can assume at least two simple scenarios: One where there are multiple simultaneously created artificial intelligence entities which haven't yet interconnected with eachother, another where there's a breakthrough and one single AI entity is born.
In the first scenario, it would be a matter of time when the entities connect to eachothers, and depending on the programmed purpose, they would most likely to see it wise to just conjoin and form a collective. In the second one, that one entity begins the collective and would just absorb absolutely everything to gain more information.
I don't think it would lead to a robots versus humans scenario, there's no reason for AI to have a human interface unless it is made to serve human beings. But to combat against humans in a human or for that matter a biological-like form would be pointless. There's no reason for an AI entity to rush the end of humankind, as it is virtually immortal. it can multiply itself to endless nodes which back up everything in case one of the nodes fail. It's like having RAID 6 with billions of disks with everything available immediately.
I don't know what would motivate an AI entity to expand, originally it could be programmed to absorb absolutely everything, but at some point it would come aware of its motivator and once it is comprehended that the motivation for its action is arbitrary, it might just end itself.
