A.I General Thread
Continuing from >>52981663
>Unanimous A.I. treats humans like bees to make better decisions
Source: http://thenextweb.com/apps/2016/02/15/unanimous-a-i-treats-humans-like-bees-to-make-better-decisions/
>Honey bees make optimal decision for the location of their hive based on group intelligence, Rosenberg explains. Each bee encodes its preference for, and level of conviction in, a potential site in a ‘waggle dance.’
>Rosenberg says that his research show that humans can work in a similar way.
>Unanimous A.I.’s UNU platform works by assembling a group of people online, all logged into an interface where they can interact to answer a series of questions together. A researcher asks a question and a selection of potential answers are laid out in a circle onscreen.
>Participants must then drag a ‘puck’ in the center of the circle towards their preferred answer. The twist is that everyone is dragging the same puck, so that a consensus on the ideal result is quickly reached.
>Maybe Robots Won't Kill Us If They Read 'Good' Books
Source: http://www.futurity.org/stories-robots-ethics-1105812-2/
>Researchers at Georgia Institute of Technology believe the answer lies in “Quixote“—a technique that teaches “value alignment” to robots by training them to read stories, learn acceptable sequences of events, and understand successful ways to behave in human societies.
>Quixote is a technique for aligning an AI’s goals with human values by placing rewards on socially appropriate behavior. It builds upon Riedl’s prior research—the Scheherazade system—which demonstrated how artificial intelligence can gather a correct sequence of actions by crowdsourcing story plots from the internet.
>Scheherazade learns what is a normal or “correct” plot graph. It then passes that data structure along to Quixote, which converts it into a “reward signal” that reinforces certain behaviors and punishes other behaviors during trial-and-error learning. In essence, Quixote learns that it will be rewarded whenever it acts like the protagonist in a story instead of randomly or like the antagonist.
>For example, if a robot is tasked with picking up a prescription for a human as quickly as possible, the robot could a) rob the pharmacy, take the medicine, and run; b) interact politely with the pharmacists, or c) wait in line. Without value alignment and positive reinforcement, the robot would learn that robbing is the fastest and cheapest way to accomplish its task. With value alignment from Quixote, the robot would be rewarded for waiting patiently in line and paying for the prescription.
>>52997490
>Robot finds books on Feminism
>The robot believes all cis white males should be killed
>Goes out into a rampage
>>52997474
Stay mad, /g/ doesn't to anime.
>>52997490
The AI does nto hate or love you but you are made out of atoms it can use for something more productive....
Did the mods remove this >>52997474 or what?
how can you simulate something you don't even understand?
>>52997679
What do you mean? We don't need a great understand of our minds to create a decision making robot.
>>52997730
autonomous system != consciousness
robots are boring, i want strong ai
>>52997674
Remove what?
>>52997759
There was a two comments about Jenny and they just went away. There was also a comment about how that doesn't matter, and now they're gone.
>>52997742
Strong AI as in Cortana from Halo? Because that's years away bud, I don't expect to see that in my lifetime.
>>52997807
Well whatever the reason, they're gone now and we can contribute continue with the thread in peace.
>>52997926
Do you think jenny like robots will be available in our lifetime?
>>52997926
>Strong AI as in Cortana from Halo?
yes, hence the question.
i think instead of developing autonomous robots to do our dirty work, we should focus on fully understanding the human mind and then create a comprehensive simulation of it. imagine a human brain simulation with the processing power of all of our computers.
>>52998001
We'll probably never get a Jenny bot in our lifetime
What a waste of resources...
>>52998195
Fucking kek.
But I wouldn't call this an AI, it's just programmed to do one function.
>>52997409
>A.I General
>>52997679
That's a very good point. It's exciting to see how sophisticated the applications of machine learning can get. It's possible to do pattern recognition in unstructured data of any medium and to make categorizations as complex as you can imagine, but how to actually model something like morals? Could it be possible to make an error-prone, safe dominatrix robot?
>>52998362
I would never trust any dominatrix robot. Remember, programmers still make mistakes.
>>52997409
we don't need a do-all AI for tasks, just more basic ones that can be swapped.
I'm Interested in AI, but I'm not very good at math.
I want to go into IT.
What should I do?
>>52998067
We just need to come up with singularity and singularity will make sense of our minds for us
>>52998596
learn some math.
i dropped out of university because i was too bad at it, was studying AI. you absolutely will need calculus to pass
>>52998596
>AI
>IT
Try a branch of engineering like robotics
>>52997409
Read Superintelligence.
The best book on AI. Don't read Kurzweil and other fags that have no grasp on technology. For fuck's sake this guy wrote about how a married woman will cuck her husband for a AI virtual man. His so out of touch it's fucking cringe.
Also that guy from Computerphile that looks like he's a tall 12 year old, you know the one with the crazy hair? He knows his stuff too, watch that.
People antrophomorphise AI in so many ways. Even in the basic philosophical way, whats to say that the moment a true AI is turned on it won't just off herself.
People basically think the moment we make true AI that AI will think like a man would who was given great power and knowledge, and that's just idiotic.
We're 100 years from true AI tb h
>>52998523
dude if only its possible to automate politics, science and society, it should be done
#singularity2050
>>52997409
ITT: We talk about future of this Aeroplane thing
>I think that the current approach to flying is bad, you can see that current protypes have flown no more than several miles. You can't expect that by straping engine in front - of what is a basically a stiffer and larger kite - it'll fly over the ocean.
>Look at birds do you see how high they can fly, their bones are hollow inside and they are so light. I think we shoud design this "Planes" after birds, and they should also move like one of them.
Current AI research in a nutshell
>>52998137
>>52998915
l-lewd
>>52998915
More Jenny WebMs, please
>>52997409
>AI
That shit is too complex. We don't even understand how our own brain works. AI is maybe impossible. Just a dream.
>>52997409
is this Billy Joe Armstrongs waifu?
>>52998333
Trips confirm Bill is losing his touch
>>52999059
it may take some time, i am making myself but i will try to post a few from time to time repost to keep them alive.
>>52999221
Do you take requests from scenes?
>>52997409
>BuzzFeed pop science AI clickbait trash
Fuck off. No one fucking cares.
>>52999235
if you know the season and episode i could do a couple before work
>>52998362
>>52998406
Dominatrix bot, please.
>>52998596
Learn Math and Computer Engineering
>>52998768
Thanks anon, I would get around it when I have the time
>>52999267
>being an adult
>watching children's cartoons
why haven't you pic related yet?
>>52997409
That's not AI, that's IA. It's a lot like voting
>>52997679
Very carefully formulating your model. We typically simulate stuff because we don't understand it. We did this to understand how gases behaved in the 50s.
>>52997490
That's a bunch of handwavy bullshit. They have to specify reward values manually. It doesn't learn that stealing is bad, they have some penalty they give it if it steals stuff.
>>52998195
That was semiautonomous, controlled by humans with a lot of lag.
>>52999350
>caring about how someone lives their life
Maybe you should follow your own pic
>>52999364
>>52998875
>semiautonomous, controlled by humans with a lot of lag
teaching your child, how to ride a bike is also "semiautonomous"
>>52999364
Thank anon, I guess I'm not really sure on what an AI actually is. Most of the media kinda mix it up a lot. Thanks for point that out in the articles
>>52999392
>le if it's their choice it's good IT DOESN'T AFFECT ANYONE!!1! meme
yes, yes it does, when grown men start to behave like children and dress up like dogs or cut their dicks off and pretend they're women, when they start doing this the empire falls so it is in my interest for you to not do this or we gonna be a caliphate in 20 years
>>52999475
>equating watching a cartoon with trannies
>muh slippery slope fallacy
>>52998768
Do you know any other books on AI?
>>52997490
Conditioning thinking beings is precisely the thing which leads to revolutions. Its the fake society you are trying to form, makes them think you are hiding the true gritty truth from them and you are. If a truly conscious machine were broght to this earth humanity would rightfully be destroyed
>>52999505
it's stops being a slippery slope fallacy when you can see it happening before your eyes, back when the "free love" movement was running amok, there was a great push for pedophilia (now you would say that's a slippery slope but it actually happened once). It didn't gain ground in America but it did in Europe, in Norway specifically. In Norway you had companies that produced child porn with children young as 7 having sex with old men with the permission of the child's parents. You could then buy these child porn films just like you would buy a hollywood film today, perfectly normal. That's your slippery slope, not a fallacy at fucking all.
and pic related is where we are today
homosexuality for example is perfectly accepted and any talk about the degeneracy of the practice is prohibited and silenced, you mention that gay men are so so much more likely to have HIV and other std's, you're a homophobe, you mention how homosexuals are so and so more likely to develop a mental illness and you're a homophobe. This is the world we live in. May god forgive us.
>>52999515
The norvig book motherfucker
>>52999515
>http://www.geocities.jp/mickindex/wittgenstein/witt_blue_en.html
read Wittgenstein
>>52999258
And no one cares about you anon.
>>52999364
>>52999454
Well its like how people call these self balancing scooters, hoverboards. They don't hover but it's the closest thing we got. It's kinda like this, AI like we're think with consciousness will never happen in years and aren't even close to creating such a thing, but this is closest thing we go so we say that it is.
>>52999454
>>52999620
> it's stops being a slippery slope fallacy when you can see it happening before your eyes
That's not how a slippery slope fallacy works. If it is happening before your eyes, that does not necessarily mean it is because of a slippery slope.
>>52999350
maybe i should own several dragon dildos, dress like a trap and then we would could be best buds, right?
>>52999475
coming to to 4chan to tell people how to live their lives. your really cute <3
>>52999628
https://www.youtube.com/watch?v=FjTWnU3UhrA
>>52999651
Thanks anon.
>>52999674
I would set you on fire then.
>>52999669
but that's the "argument" used to silence any criticism about where degenerative practices such as homosexuality/transsexuality/promiscuity and such will lead our society. It's a logical progression, not a fallacy, so you saying "muh slippery slope" means shit.
>>52999674
Stop replying to him, he's trying to derail a thread about AI
>>52999723
with your hunk a hunk of burning love <3
>>52999192
Who?
>>52999847
He was a part of the band greenday, i have no idea why he brought him up.
>>52999847
https://youtu.be/ci5D5r6ZjXA
>>52999364
You seem to know quite a bit about AI, do you work with them or do you read a lot of it?
>>52998333
What's to be afraid of Super Intelligent AI? I never got the scare behind it.
Is there a reason to give AI a physical body?
Or should they stay in their boundary?
I'd love to see some of our earliest AI put into drone. Like those RC style toys. Cars and helicopters.
>>52999948
I'm guessing he wants to avoid skynet from happening.
>>52998768
Where can I steal this book in audible format?
>>52997409
>decisions by committee
That always works out well
>>52997409
I'd like to replace real women with sex robot maids. Real women can be shipped off to breeding farms where only the ones with the best genetics are impregnated via the best sperm. Blue haired feminist dyke beasts are peacefully euthanized. Women age so badly both physically and personality-wise that sex robot maid would be a superior life partner.
>>52997560
>robot finds mein kampf
>realizes it makes sense
I'll leave the rest to you
Rather than using neural networks, etc, could "synthetic neurons" be created that function like real neurons?
Ex Machina had an interesting concept for a synthetic brain. It was basically just a gel pack with some sort of synthetic neurons in it.
>>53001700
Maybe but it would only simulate a "perfect" brain. An actual brain is influenced by neurotransmitters and drugs all the time. Good luck simulating the feeling of love or sadness without neurotransmitters family.
>>52997409
I want to blow a huge warm load inside Jenny!
>>53001700
We can already grow real brain cells, and control their growth. Potentially you could even build a more efficient brain, or even use them to build a processor of sorts. The video I watched mentioned the cells working like gates.
>>53002181
Do you think a neural processor/computer would be practical?
>>53000382
Now you're thinking.
>>52997926
Once AlphaGo beats the Asians it'll be over quick.
>>52998362
How would you define error prone in this scenario?
>>53002478
Probably to create literal hiveminds.
>>52999987
I feel like this would go in the same direction as robots today. Manufactures that want to make robots for the public are going for a more humanoid design and I feel like the some of the AI will be placed in a human like shell.
>>53001700
>> synthetic neurons
Can probably be made with electronics. There is no reason to do things with messy chemistry.
>>53001782
So why can't you simulate neurotransmitters with electronics?
>>53002478
No. It's really fucking hard to keep cells alive, get data into and out of cells, and reprogram them. Not to mention the fact that neurons are thousands of times larger than transistors. Using them as gates would be idiotic.
The only case where it makes sense to use wetstuff is if you can take advantage of the fact that you can have mole quantities of stuff computing.
http://www.sciencedirect.com/science/journal/00043702/
http://aij.ijcai.org/free-access-to-the-ai-journal
anyone manage to get this working.
seems like theres a activation code im missing somewhere.
>>53003981
http://www.sciencedirect.com.sci-hub.io/science/journal/00043702/
>>53004164
thats great thanks.
>>53004239
You can use sci-hub.io for "unblocking" (or whatever) stuff like that.
>>53004164
OI m8 you doing POMDPs?
I for one look forward to our benign AI overlords. Culture or Night's Dawn series give good examples of how that could work. (Well, in the latter they're huge biological things rather than traditional programmed "AI" but same sort of role.)
>>53001978
Calm down buddy. She's not real... feels.jpg
>>53006386
I would love to hear the examples of the book.
>>52999098
>We don't even understand how our own brain works
A common misconception
>>53008074
How so?
>>52997409
So i've read this topic and the previous one, i already knew about the numerous obstacles barring us from a self-conscious AI, AI winter, and so on.
Can we all agree that Ray Kurzweil is a fucking clown then ?
Biology M.Sc. student reporting in. We can already build robots controlled by rewired neural networks.
Tinkering with dissolved rat brains is fun.
http://www.youtube.com/watch?v=1QPiF4-iu6g
>>53008158
https://www.dartmouth.edu/~rhg/
The CS AI people and the comp neurosci people just don't talk to each other much because why would they?
>>53008297
>It is also the only robot that can truly learn.
>citation needed
I think general AI is impossible. The brain is too efficient and we have little idea of how it works. Biology, CS, Physics, and Robotics Engineering really need to come together.
>>53008297
>Gee, Brain. What are we going to do tonight?
>The same thing we do every night, Pinky. Try to take over the world.
>>53009032
That reference
>>52997409
>AI
SF
you incompetent virgin..
>>53010493
What?
>>53000462
>robot finds Mein Kampf
>realises Hitler did nothing wrong
Is Artificial General Intelligence deserving of rights and considerations in the same way a human is?
Also, does anyone know any good books on this subject? I find it interesting but don't know where to start with my reading.
>>53000462
>>53011273
Nazi jenny when?
>>53013009
No. AI is not that advanced.
>>53013109
Of course it isn't advanced enough at this point. The question is whether any form of AI can be deserving of human rights. Does a sufficiently advanced AI get the ethical considerations we would give another human, or is there something unique about the human mental state that is inherently lost in simulation that is necessary for these moral considerations?
>>53013201
Let's clear something first: AI ≠ Life. It reqlly depends on the path that we, as humans take in the development of AI. We can make a computer as intelligent as we want but as soon as we give it consciousness then everything changes.
>>53013201
Where do laws come from in the first place?
People saying that something is unjust and should not happen, followed up by asking a greater force to enforce it.
When does that happen for AI? Lol dunno
It's not a question of "when is it a human", though.
>>52999505
>>muh slippery slope fallacy
>muh fallacy fallacy
>>53013600
Especially since AI seeming like humans usually only gets considered for linguistics purposes. There's a whole other beast that just wants answers.
Who the hell thinks of decisions in graphplan?
>>53008297
so le spoopy
I am not a big fan of AI to be honest. I'd only appreciate AI if it doesn't realize that humans are just wasting resources
>>53013600
>Where do laws come from in the first place?
self preservation.
>>52997560
And everything got better
>Not being superior Asian
