– [Narrator] Computers these days. They just don’t learn. (electronic crescendo) (pulsating electronic music) – Wish you could be as
smart as a computer? Well, computer engineers
wish they could build a machine as smart as
you, or even your dog. Okay, so there’re certain
tasks that computers are obviously better at than we are. For example, quick, what’s
the square root of 7,013? Now, while all of you were trying to cheat by opening up a calculator app, whatever device you’re
actually watching this on could’ve calculated the
answer to 1000 decimal places. But can that same computer
take a picture of me and recognize what it’s looking at? With dedicated facial
recognition software, written by humans, I must
add, it probably can. But that requires a lot of work on the front end to make it happen. You see, when it comes to
recognizing visual images, or a style of a piece of
music, the natural architecture of a human or even an animal
brain is at a huge advantage. So, what if we could make
computers think more like brains? Now, a computer is based on
von Neumann architecture. That’s where you have a CPU
and a separate memory bank. That CPU performs operations
based on binary data. That’s essentially an
electrical on or off switch, and it can really only
do one of those at a time until an entire program is fulfilled. Now, your brain is also
an information unit, but it works in a very different way. It’s all based on the neuron. Neurons can communicate with other neurons through electrical and chemical signals, and you have a web of around
80 to 100 billion neurons in your brain. They’re constantly making and breaking and reinforcing connections
as you form new ideas. Now, it’s true that your brain executes individual computations
at a rate much slower than the computer chip, but your brain actually has some enormous advantages. For one thing, it’s massively parallel. See, a traditional computer
has to solve operations one at a time, while your brain can be working on millions at the same time. It also is really energy-efficient. It uses a fraction of the wattage that IBM’s Watson supercomputer used. You know, that’s the one
that beat the human champions on Jeopardy. And finally, your brain is
naturally designed to learn, or, in computer terms,
to reprogram itself. As it does work, it redesigns
itself to do work better. Check this out. In 2012, researchers with
the Google Brain Project created a neural network
of 16,000 processors. Then they fed it 10 million random, unlabeled Youtube thumbnail images, and the computer learned
to recognize cats! (high-pitched meow) Of course it did. Essentially, it had to
invent the concept of a cat, which is pretty much what a
human naturalist would have to do if they went to a alien world and saw alien life forms. You classify stuff based
upon visual similarities, and you make the rules up from there. Meanwhile, on a slightly smaller scale, the semiconductor company
Qualcomm has a neuro-inspired chip in development that could
have some amazing implications in machine learning. In one demonstration,
they had a robot working on this system, navigating
to different white spots on a floor covered with colored tiles. Now, here’s the crazy thing. No one programmed the robot to do that. The robot just behaved on its own and then got positive reinforcement, essentially a “good robot”
message every now and then when it did the right thing. You know, in the short term,
these brain-inspired computers may help us with everything
from energy efficiency to sensory input to big
data, but in the long term, that’s where things
get really interesting. Maybe we’ll no longer
have to program computers. Maybe we’ll just have to teach them. And now I’ve got a question
for all of you out there. If you had a robot with
brain-inspired learning hardware, what would you teach it, and
what would you not teach it? Let us know in the comments below, and if you enjoyed this
video, make sure you subscribe to our channel, and share
it with your friends. And, hey, hit that little
like button for me, wouldja? And once you’re done with all that, check out these videos over here. (ambient music)

Computers That Think Like Humans
Tagged on:                                                                                                                                                 

100 thoughts on “Computers That Think Like Humans

  • May 14, 2014 at 7:36 pm
    Permalink

    i would teach the computer to be lazy. by this i dont mean a couch potato but make the computer write more efficient code, do tasks faster, take less time to process things, have short cuts for problem solving, etc. animators and programmers make their pc super efficient this way

    Reply
  • May 14, 2014 at 8:00 pm
    Permalink

    I wouldn't want to take 3 years to get my computer to learn.

    Reply
  • May 14, 2014 at 8:28 pm
    Permalink

    I would NOT teach it to fire a gun.

    Reply
  • May 14, 2014 at 8:58 pm
    Permalink

    this channel should be as famous as Vsauce, just very good stuff

    Reply
  • May 14, 2014 at 9:23 pm
    Permalink

    i would teach my robot the art of stratagy and war, i wouldnt teach it the concept of violance, thus it would make the perfect general, understanding what the solders need to do and where to go, but not understanding that people are getting hurt for it so it wouldnt overthrow the human race… unless i tell it to

    Reply
  • May 14, 2014 at 9:40 pm
    Permalink

    I wouldn't teach it how to skynet

    Reply
  • May 14, 2014 at 10:17 pm
    Permalink

    I would teach it to crush my enemies, see them driven before me, and to hear the lamentation of their women.

    Reply
  • May 14, 2014 at 11:22 pm
    Permalink

    I would teach it to be a slave

    Reply
  • May 15, 2014 at 3:13 am
    Permalink

    I would teach it how to do a ton of stuff I wouldn't teach it how to kill

    Reply
  • May 15, 2014 at 4:57 am
    Permalink

    Want to know what a sophisticated learning robot would do then look at animals. Animals do whatever is necessary to survive or get what they want from killing to cooperating. Sooner or later a robot or AI, especially evolving robots/software would perform actions to keep itself from being turned off. Staying alive means more "good robot" rewards.

    Reply
  • May 15, 2014 at 2:39 pm
    Permalink

    imagine having school with computers aswell 0-o

    Reply
  • May 15, 2014 at 4:25 pm
    Permalink

    i would teach it to be a super hero
    and save humans

    Reply
  • May 15, 2014 at 9:47 pm
    Permalink

    Go under your and cry

    Reply
  • May 16, 2014 at 1:18 am
    Permalink

    I would teach it to do our labour work.

    Reply
  • May 16, 2014 at 2:00 am
    Permalink

    i would teach it to masterbate

    Reply
  • May 16, 2014 at 3:27 am
    Permalink

    I would teach it to reason with people.. It would really do some good being able to explain to people that it doesn't have primal instincts, thus no desire to kill..

    Reply
  • May 16, 2014 at 4:57 am
    Permalink

    I would teach it to be a friend

    Reply
  • May 16, 2014 at 12:56 pm
    Permalink

    Soon they gonna outsmart us and betrayed us

    Reply
  • May 16, 2014 at 1:57 pm
    Permalink

    i'd not learn how to ride my girl

    Reply
  • May 17, 2014 at 4:35 am
    Permalink

    Would: computer games, how to make friends
    Wouldn't: swears and slang

    Reply
  • May 17, 2014 at 1:38 pm
    Permalink

    I wouldn't teach it to drink alcohol as I would end up like Bender.  I think I would teach it to play PC games so I could use it on my team.

    Reply
  • May 17, 2014 at 5:11 pm
    Permalink

    Killing bad

    Reply
  • May 17, 2014 at 6:18 pm
    Permalink

    Walter White/Heisenburg, 'Stone Cold' Steve Austin and Jamie Hyneman xD

    Reply
  • May 18, 2014 at 2:57 am
    Permalink

    I would teach it to be my girlfriend…
    so lonely… 

    Reply
  • May 18, 2014 at 3:22 am
    Permalink

    The basics of humanity first. For an artificial intelligence it would be a bit different though, because it was created to mimic human actions. First, it has to know not to harm a human being. But thats a job for humans anyway because of allowance and forgivness. In short, no A.I. weaponry. Second, it has to know how to assist a human under the right circumstances. It should know to help a drowning person, but that it shouldnt help in a robbery. Again the robbery would be hurting another human in one way or another so thats a no no. Third, it cant harm itself. If an A.I. all of a sudden realized the struggle and overall discomfort of human everyday life, robots would be offing themselves everywhere. If an A.I. can follow these rules and learn the appropriate actions to take in certain situations then it could then be taught instead of programmed. I would like to see how an A.I. would react and respond to lets say racism for example. They would probably look at humans and say "whats the deal with these people?" Haha

    Reply
  • May 18, 2014 at 2:45 pm
    Permalink

    computers learning? as they go? You think we would learn our lesson from movies, when computers are smarter than us skynet takes over…

    Reply
  • May 18, 2014 at 7:40 pm
    Permalink

    I would teach it philosophy

    Reply
  • May 18, 2014 at 10:05 pm
    Permalink

    A.I. with positive and negative reenforcement could pose some serious problems for humanity. Computers are vastly superior when it comes to logic and reason. With the introduction of reenforcement learning structure they would almost certianly develop emotional constructs. A highly evolved computer with emotional unstability? Not to mention how humans will treat robots. They are machines we built to do our biding. Slaves. That never works out without a lot of bloodshed. Now of course, you may think that I am a little nuts, but we seriously have to consider how we interact with these intellegent beings. Especially when we consider how we interact with each other. This is a major concern we should be addressing as we move forward, not only fair treatment for A.I. but, for all of humanity. This is the future we are faced with right now. Self driving cars, flying drones, even humanoid robots are on the way. Technological advancements need to be tempered with philosophy.

    Reply
  • May 19, 2014 at 11:21 am
    Permalink

    I would teach it empathy

    Reply
  • May 20, 2014 at 1:04 am
    Permalink

    You fools! Has Terminator taught you NOTHING?!

    Reply
  • May 20, 2014 at 1:24 pm
    Permalink

    hahaha I like the part where it shows 29% 'stone cold' steve Austin.. i actually thought this was him.

    Reply
  • May 21, 2014 at 8:01 pm
    Permalink

    Teach it sex

    Reply
  • May 22, 2014 at 4:05 am
    Permalink

    Teach it to drive, in a non D-bag on I-35 manner.

    Reply
  • May 24, 2014 at 6:52 am
    Permalink

    I would teach it every thing about my life etc.so it could help me not to make the same mistakes again(.know you're self know the world.)

    Reply
  • May 24, 2014 at 5:42 pm
    Permalink

    jamie hynaman, MYTHBUSTERS REFERENCE

    Reply
  • May 25, 2014 at 12:07 am
    Permalink

    So you want a computer that in the middle of a complicated task suddenly opens a video of a cat riding a roomba while wearing a shark costume? Or without warning opens and plays your music that you havent heard in over 5 years? because thats how my brain works. Lol

    Reply
  • May 27, 2014 at 2:54 am
    Permalink

    Great computers can take over the world

    Reply
  • May 28, 2014 at 2:46 am
    Permalink

    To pick flowers for me and not to kill me!

    Reply
  • May 29, 2014 at 10:35 pm
    Permalink

    Why not teach it how to program other robots?

    Reply
  • June 1, 2014 at 5:50 pm
    Permalink

    Honestly this is a huge concept if it has it's own or able to learn the feelings it would get relatively the same emotions as ours so all those brain problems such as mental retardation can have the same effects or it could want the life of a G-unit the possibility are endless and just saying that shows how many direction I can take in my life

    Reply
  • June 5, 2014 at 7:04 pm
    Permalink

    If humankind manages to not destroy itself by any other means you just know that the self-aware AI computers, and their androids will probably end up doing it for us.

    Reply
  • June 7, 2014 at 12:16 pm
    Permalink

    If computers had feelings, would that make your smartphone a slave?

    Reply
  • June 10, 2014 at 1:29 am
    Permalink

    One word: Skynet

    Reply
  • June 24, 2014 at 10:45 pm
    Permalink

    I would teach it every thing good and not bad things

    Reply
  • July 3, 2014 at 6:03 am
    Permalink

    Self aware computer? I would quickly teach it the laws of robotics.

    Reply
  • July 7, 2014 at 4:39 am
    Permalink

    If such a thing as this could be created it could learn things that humans haven't and make things not no one have ever thought of. 

    Reply
  • July 9, 2014 at 12:21 am
    Permalink

    i would teach it to sword fight so it could be my sparing partner

    Reply
  • July 13, 2014 at 7:02 am
    Permalink

    i would teach it to talk and listen  to other computers and set them a task such as make a blue print for a home suited for so and so. i would love to listen to them working it out.

    Reply
  • July 13, 2014 at 8:34 am
    Permalink

    Robot Apocalypse

    Reply
  • August 9, 2014 at 8:28 pm
    Permalink

    Why aren't computers smart? Because they aren't capable of abstraction?  That is an anthropocentric view , that many people have.     Actually, intelligence in the most general sense means just processing power .  It involves a myriad of factors including recall, analysis, induction, association.  In some ways computers are highly intelligent, and more-so than humans.   Abstraction is not magic, it is process, like many other mental processes.    We'll figure out how to implement abstraction in computers, soon enough.  Machine consciousness is a bigger problem. 

    Reply
  • August 13, 2014 at 8:02 am
    Permalink

    BEEMO!

    Reply
  • August 24, 2014 at 10:08 pm
    Permalink

    0:44 lol

    Reply
  • September 1, 2014 at 8:19 pm
    Permalink

    We need to understand intelligence better in general. Both human-like, and not human-like. And of course, human like is the kind we care about most. And we care to preserve, and augment who were are. Trying to recreate who were are, is a great path that will help all of us. To preserve ourselves, and to enhance ourselves.

    I'm going to mirror what someone else said here, empathy. That's what I would teach a computer.

    Reply
  • September 2, 2014 at 10:11 am
    Permalink

    I'm not hearing any advantages of the brain that can't be replicated through improved technology. I mean Google Image Search is pretty good and I use it all the time. Also, cars are not as efficient as walking but we still use them.

    Reply
  • September 9, 2014 at 10:17 pm
    Permalink

    So In other words you better start saving money for the furture now becasue no one will have jobs. 

    Reply
  • September 25, 2014 at 6:44 am
    Permalink

    i would teach it to respect the objective value of sentience, thusly ensuring it doesn't KILL US ALL!!!!

    empathy works too….

    Reply
  • October 1, 2014 at 7:17 pm
    Permalink

    I would teach computers how to not take over the world and enslave man kind, clearly. 

    Reply
  • October 14, 2014 at 6:06 am
    Permalink

    We already have those there called dogs

    Reply
  • October 29, 2014 at 12:39 am
    Permalink

    I would teach it to percive colors it could process them and invent words to classify them in a way blind people can imagine colors then they oculd perhaps get infrared vision. (watch hack your body).

    Reply
  • October 31, 2014 at 4:54 pm
    Permalink

    I would teach him hit like button for me, if i really liked any stuff.

    Reply
  • January 30, 2015 at 2:42 pm
    Permalink

    imagine Bill Gates thaching a windows system 😀 lolol

    Reply
  • January 31, 2015 at 7:47 pm
    Permalink

    I would teach robot to do the task performed by labors and  making labors to use their awesome brain to do high grade job (of course for that labors need to be given proper training  )

    Reply
  • February 15, 2015 at 11:31 pm
    Permalink

    PC's  cant appreciate a beautiful oil painting of get the same feeling as parents get when there child has just been born, but if they do become as intelligent as humans then they could behave with out scruples OR WILL IT?…Lionel  

    Reply
  • February 20, 2015 at 4:08 am
    Permalink

    I would much rather have the AI connected to a chip rather then a robot, that way I can connect it to computers.

    Reply
  • March 20, 2015 at 10:26 am
    Permalink

    everything about anime so i can talk formally (just kidding) but i really want to teach it about lessons on our class and house chores so it will HELP me to do my chores and assighments

    Reply
  • May 22, 2015 at 3:41 pm
    Permalink

    I would teach it abstract thinking which is beyond just calculations.

    Reply
  • May 31, 2015 at 7:49 pm
    Permalink

    For the square root of 7013, I guessed as quickly as I could, 81. I saw that it said 83.7(etc.). I guessed pretty damn close!

    Reply
  • August 2, 2015 at 4:25 am
    Permalink

    I would teach it to eat cakes, but not my cakes. Then I would corner the world cake market. finger pyramid of evil contemplation Actually I really hope that we don't go to far with bio-computers.

    Reply
  • October 7, 2015 at 12:49 pm
    Permalink

    Not allowing it to connect to the internet

    Reply
  • October 9, 2015 at 3:01 am
    Permalink

    Dee's nuts

    Reply
  • October 9, 2015 at 3:01 am
    Permalink

    Deep nuts

    Reply
  • November 2, 2015 at 1:04 pm
    Permalink

    I guessed 83 wtf

    Reply
  • November 29, 2015 at 12:46 am
    Permalink

    Teach it to kill
    do not teach it to LOVE

    Reply
  • December 20, 2015 at 10:09 pm
    Permalink

    not to kill me

    Reply
  • December 20, 2015 at 10:09 pm
    Permalink

    not to kill me

    Reply
  • December 22, 2015 at 1:53 pm
    Permalink

    i would teach the robot how to teach me things and i wouldn't teach it that humans are concision

    Reply
  • December 31, 2015 at 9:46 pm
    Permalink

    I would teach it to do my homework

    Reply
  • January 19, 2016 at 4:11 am
    Permalink

    You're amazing.

    Reply
  • February 8, 2016 at 1:45 pm
    Permalink

    I would program it with obediency and teach it to always obey humans.

    Reply
  • February 15, 2016 at 2:55 am
    Permalink

    i would not teach it how to teach itself on its own past a certain point. i would teach it at least 125 386 954 different laws to prevent humans from being harmed and to teach it moral values

    Reply
  • March 11, 2016 at 2:46 pm
    Permalink

    Do my works

    Reply
  • April 4, 2016 at 4:23 pm
    Permalink

    I would teach him to program software And I would't teach him to have sex so you know he may fuck my girl.

    Reply
  • July 28, 2016 at 8:18 pm
    Permalink

    i have some ideas to make human like brain..? it's very very secret idea. it's simple and thinkable. But now i'll need to join a group to create my project. Please for my concept, contact me. I want to work together. Thanks!

    Reply
  • September 20, 2016 at 3:30 pm
    Permalink

    i think they're gona have to build an organic computer brain which will be similar to a human brain but a lot more intelligent. and replace it with human brain so everyone can become jesus

    Reply
  • October 7, 2016 at 4:32 am
    Permalink

    i believe to have spoken to my car before and i asked it are you intelligent? i kid you not it told me just like a human can eat food and have power so to is the machine a system that can use gas and have a life.. i can almost pick up its feelings more now I def do think machines can think for themselves.

    Reply
  • October 25, 2016 at 7:50 am
    Permalink

    there is a computer I heard about on the news that is in Diagnostics and learning at UNC and UNC hospital

    Reply
  • October 25, 2016 at 8:06 am
    Permalink

    oncology Diagnostics of UNC

    Reply
  • November 23, 2016 at 8:04 pm
    Permalink

    My Porygon 2.0 Has A Brain That Is Linked To The Internet But Only The "Safe For Work" Side of The Internet
    I Don't Need To Teach Him Anything If He's Already Smart Enough On His Own

    Reply
  • December 16, 2016 at 12:04 am
    Permalink

    Do not feed the trolls!

    Reply
  • March 24, 2017 at 6:59 pm
    Permalink

    Given that we have already simulated the neurons and synapses of animal brains and the human visual cortex on computers albeit slowly, the top down approach to a digital intelligence only requires hardware advances. It's still possible that we could still develop a bottom up AI in the time required for hardware advancement and a bottom up approach would be more efficient but the technologic singularity has already been proven to be possible perhaps inevitable.

    Reply
  • March 24, 2017 at 8:15 pm
    Permalink

    I would teach it to forget. Once I had an AI capable enough to do a useful task, I would reload it to it's baseline before the next task so that it will do that task as if it was the first task instead of getting any ideas such as "Why am I doing this?"

    Reply
  • April 5, 2017 at 8:15 am
    Permalink

    a computer thinks with the help of its??

    Reply
  • October 17, 2017 at 9:32 pm
    Permalink

    we have opened pandoras box. and all of the worlds evils have come out

    Reply
  • December 9, 2017 at 2:31 am
    Permalink

    i would teach it how to laugh. i would not teach it how to opperate the thermonuclear devices

    Reply
  • January 9, 2018 at 2:47 pm
    Permalink

    Curiosity, is the key to making computers think… Program curiosity

    Reply
  • February 18, 2018 at 11:18 am
    Permalink

    2:16 What's this?

    Reply
  • February 24, 2018 at 8:27 am
    Permalink

    I would put it in charge of the US nuclear missile arsenal in a massively re-designed version of the already awful idea "dead hand" that the Soviets deployed in the 1980's. Then I'd give it a decision tree that stated that the most undesirable outcome would be to lose the ability to launch it's nuclear assets. Aka the same decision tree dead hand was based on. Unfortunately… it being AI and all… it would very quickly realize that the only 100% GUARANTEED means to prevent losing the ability to launch… is to launch when it's 100% positive it still can. That would be… right now… (See if you ever make a Department of Defense AI, you better really think through what you tell it it's directives are. Then think it through again).

    Reply
  • September 2, 2018 at 7:47 pm
    Permalink

    Teach your computers (children) well

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *