Holy Grail of AI (Artificial Intelligence) - Computerphile
Вставка
- Опубліковано 23 сер 2024
- Audible free book: www.audible.com...
Why can't artificial intelligence do what humans can? Rob Miles talks about generality in intelligence.
Sean Comments/Questions (For those who can't hear him clearly)
11secs: "This was the hill climbing algorithm?"
2min 40sec: "recently with Professor Brailsford we did the idea of The Turing Test, so that strikes me from what you're saying that that's a very specific domain: pretending to be a human talking?"
4min 2sec: "is that like 'humans have been changing the world to meet their needs?' "
4min 40sec: "but on a bigger scale, as you say on a grander scale, building a dam, and er irrigating a field, and putting a pipe to your house and allowing you to have a tap {fawcett} is doing the same thing but on a grander scale."
6min 39sec: "all these dimensions, if you try to brute force infinite dimensions you're gonna fall over pretty quickly?"
6min 45sec: "change the world" (in ref to him picking up the drink!)
Retro Z80 Computer: • Home-Made Z80 Retro Co...
Hill Climbing Algorithm & Artificial Intelligence: • Hill Climbing Algorith...
The Turing Test: • The Turing Test - Comp...
Arduino Uno: • Arduino Hardware - Com...
Rabbits, Faces & Hyperspaces: • Rabbits, Faces & Hyper...
Thanks to Nottingham Hackspace.
/ computerphile
/ computer_phile
This video was filmed and edited by Sean Riley.
Computer Science at the University of Nottingham: bit.ly/nottscom...
Computerphile is a sister project to Brady Haran's Numberphile. More at www.bradyharan.com
Whoever added "Gulp" into the CC on this video is a saint.
More of this guy please, he's awesome!
Jack Reaper agreed, he is brilliant and I love the way he speaks.
Jack Reaper He looks like a 60's pop star! :D
Jack Reaper He is articulate in the right way
Jack Reaper He looks like, (except from the hair) and has many of the the same mannerisms, as a friend of mine. It's so strange.
Kael You have cool friends :-)
The way he described what you're doing when taking a drink was awesome!
I'm referring to 4:20 (blaze it)
Stephen Kamenar Which was what!? I seriously listened to what the cameraman was asking several times over and I couldn't tell at all! Thanks in advance :D
chemxcore I updated my comment
Stephen Kamenar OH. LOL. When he described taking a drink. I thought you meant when the guy actually took a drink of his mug. I still don't know what he said then. Something about it being threatening. BLAZE IT LOL, love that
chemxcore Oh, I see. I didn't even realize he took an actual drink. How intelligent of him.
yep noticed it too, and i got triple coincidence, i watched the video high, it was around 4:20 when it got to that part of the video, and at that 4:20 timing in the video he talks about picking a glass and drinking water, the most typical problem when high, and water being the most important resource for stoners after weed
The guys from Nottingham Computer Science came to my university in Campinas, Brazil, last month and told us about their AI research. I got really excited and I hope I can research with you soon!
Fuvity Hi, it was actually me giving the talk in Campinas, great to hear you enjoyed it. Following the trip I've decided to do a few computerphile videos myself to talk about some of the research we do. Hope to see you in the UK one day!
I've seen your last video hahaha
Shout out to whoever wrote the transcript for this one. Big fan of the gulp at 1:59
Just came here after about a year to thank you for the book you recommended. Superintelligence was an awesome read. Everybody (literally every person on earth) should check it out.
Don't take it too seriously. It's science fiction in disguise.
I always think that more conditions should be attached to recommendations like these. In these dark ages where students about to overcome the most prestigious high school in the country have read only one book recreationally in their whole lives, not everyone will appreciate it. Like, an example is: Everyone who has ever played a game with decent plot at all (GTA:SA level?) should play Omori.
@@recompile make a case. this is trite.
What's the authors name
@@recompile ChatGPT disagrees
> Google Car can't play chess
Google: One moment, please.
AlphaZero moment
Now is alphazero capable of driving cars is the question
This comment actually represents the problem in understanding what an actual AI is. We don't have an AI that was trained for chess and then can drive a car and, even if we could, it would (as of mid-2024) need a human choosing one algorithm over the other. In the same way, things like GPT aren't "close to AI" because they need human supervision to work. Without a prompt, the thing does nothing, because it's a tool, not an intelligent agent.
@@ekki1993 This comment actually represents the problem in comprehending comments by not reading them properly and answering impulsively instead.
Note how nowhere in my comment did I say anything about an AI. Yes, what you said is technically true. No, as far as my OC is concerned, the relevance of yours is at best tangential.
@@Anvilshock Hey, applies to you too. I wasn't talking about your knowledge of the topic, just using it as an useful metaphor. I understood your comment as a joke and I was making a tangential comment. It was tangential on purpose.
Rob is simply amazing, more videos with him!
Rob Miles is a bit like Jean-Ralphio's much smarter twin.
Rob Mckennie yes!!
He is Goooo-oooooooooooo-ooood
I fully support another commenter by saying this person should have his own channel. He's charismatic, articulate, concise, and smart.
For anyone who stumbles across this comment somehow without already knowing, he has his own channel now and it's spectacular! Plainly named Rob Miles, it's just a search away.
If you put a chess A.I. in the self driving car and the car is a mustang, It tends to move 2 lights forward and make a 90 degree turn to the right or left and goes forward 1 light. :)
😂😂😂😂😂
This was very interesting, definitely looking forward to seeing the rest of this series.
Computer proves it's human:
"I would load the board into a chess program and look up the next move"
"I don't drive. I take an Uber"
"I don't know. Jeopardy is hard"
You could rate possible AI decisions on a scale. For the water drinking example:
1 (Low). Picking up the cup and drinking the water inside.
2 (Medium). Building the dam . . . getting water out of the tap.
3 (Extreme). Realizing that water is made of hydrogen and oxygen, and gathering all hydrogen and oxygen together to have 'water'.
Rob Miles is excellent! I'd love more of him on the channel!
More videos with Rob Miles please!
Real (general-purpose) AI will have truly profound implications. If you can build an intelligence, you can also scale up your design (or just tell the machine to do it for you!). There'll be literally no work that our machines can't do for us (& better than we can).
vonkruel I'm skeptical. We're still too far away from general-purpose AI for us to really understand what its consequences or scalability will be.
3snoW I believe we'll get there in this century, and perhaps even by 2060 or so. I have no crystal ball & don't claim to know all the consequences, but it doesn't take much imagination to see that this will be a very important milestone for us. We'll certainly see the real impact of it when we get there, and I have absolutely no doubt that we will. The first real machine intelligence doesn't need to be a full human-brain simulator, but it'll probably come from an understanding of the key mechanisms in the brain that enable intelligence. As (relatively) smart as we are, it seems like this problem is tough enough that we'll have to rely on reverse-engineering to get us started.
3snoW I would recommend you to read Randall Munroe's book "What if". There in a article he discusses about the complexity and ingenuity that AI would acquire in the next couple of decades.....
vonkruel Reminds me of that Calvin and Hobbes strip;
Calvin: "If computers can think, what will people be better at than computers?"
Hobbes: "Irrational behavior."
Calvin: "Maybe they'll invent a psychotic computer."
T Landman I like your idea of ai not reaching the height of humans, but we are talking about intelligence here and when it comes to irrational behaviour it goes way off the mark of intelligence. So perhaps AI cant reach human standards or cant replicate process that happen in human brains but as Randall Munroe pointed out in his book ' What if ' , they may go far beyond us
Question, would you describe Artificial neural networks as general intelligence or would it be described as specific intelligence as it is usually implemented with a task in mind. If you consider the Hopfield algorithm, would this on it's own be an example of general intelligence before being applied and would it remain general intelligence after when it is applied to a task?
Why can't AI do what humans can? Because barely 1 human lifetime ago we were driving horse-drawn carriages down cobblestone streets and our technology hasn't had much time to develop yet.
A lifetime ago was the 40s
MilitantAntiTheist Man, you've been alive a long time then.
*****
That's why I had co-workers born in the 1920's and my great-grandmother was born in 1902 and died in 2005, the day after Christmas no less.
*****
Remember, there were people who witnessed Custer's Last Stand as children who would, later in their lives, watch the moon landing on television.
This guy is a good teacher.
3:10 Wrong. Turing's original test was composed of a series of random questions that expected intellectual answers, such as "How many items are in a dozen?" and "Given this chess game, mate me in three moves."
These AI video are magnificent. It's the only UA-cam videos that I can rewatch over and over again. Don't get me wrong, there are many good videos out there but these ones tower over them all
I wanna ruffle his hair. :3
SkrootNissu Yoctomind I wanna shave him to a skinhead.
There is a level of intimacy for everyone. You've crossed that level, haven't you?
I wholeheartedly agree
This guy is pretty cool. I was waiting for him to pop up again.
More stuff on AI? GREAT! =D
I really enjoy these vids with Rob (that's his name, right?)
2:10 "...and then we learned to become good at [driving]"
Maybe where you're from.
@@rewrose2838 pretending to not understand humor doesnt make you smart.
Aww, that snack vending machine in the background though :3
it's really great how you're trying not to have swallow noises in your videos by covering mouth or lowering volume, it's something that really bothers me in other videos when it's not eliminated
better make sure the values we give them are in line with ours.
***** Values uploaded! First question from AI: ''Where the bitches at''?
Here is a very interesting related article: "The AI Revolution: The Road to Superintelligence" (google it by name, because youtube forbids links)
It doesn't forbid links?
@@circuit10 Apparently, that was true when I posted the comment.
@@stIncMale Oh wow, I didn't notice it was 5 years ago :)
Why should the Turing test matter? If there is intelligent life in space, it might as well not be able to pass the Turing test. Consider Solaris (a book by Stanislaw Lem).
I like your references for the hearing impaired. Would have been great with subtitles for the ones who can't hear at all. I mostly read lips just in case I can't hear everything.
You could just make an AI that does 3 things. Search for new types of data, make random test spaces between known data and keep the ones with linear-ish relations.
Loved the video. Great to introduce the concepts and very efficient in relating the abstractions of the problem with it.
Actually Rob's answer to the Turing test would tell me that I'm talking to a computer. No human could know everything and part of the beauty of conversation is that an "I don't know" answer is perfectly appropriate.
What about all the self learning - deep neural networks (like the one which google allowed to play computer games and it was able to find glitches/ beat the games at record speeds, while the AI wasn't even supposed to be used on games).
+TankT9 give it access to the internet and you'll have yourself skynet in a couple of months.
***** link plz
Captcha is actually a type of Turing test, which at this time is quite effective.
And I like this guy. Brilliant.
So the general idea is you would require some type of hierarchical structure that allow you to reduce the dimensionality of the problem by selecting the required rules and boundry conditions. If you don't tell a human what his general goal is he also has no clue how to act. Give a man a set of items and ask him what to do and he won't have a clue, tell him it is a game and he will start to combine all the provided items to try to figure out in what way it can be used to play a game, using all his previous knowledge of what a game is. Tell him all the rules and objective of the game and he will try to figure out how to use those to reach the objective.
So shouldn't a general AI just have a very robust pattern recognition algorithm and huge data base in order to reduce the dimension of the problem.
does this thing look like a vehicle, then I'll apply traffic rules and try to identify what input is required to make this vehicle move. Extra layer, try to identify, by asking or observation, the region you are in, in order to find more boundry conditions in the way of laws etc.
in other words, bringing it down to cathegory of problem/goal, available tools and boundry conditions, and required input. And trying to determine all these by pattern recognition and or inquiry of other computers or humans in the surrounding.
I always feel that when people speak of human intelligence they forget to consider that it is almost completely based on experience and on what they have been told. Combining existing knowledge in existing or new ways to solve a well defined problem. If we either don't have that knowledge or the problem isn't well defined humans are pretty rubish
6:47 The scary evil laugh at the end
All these comment are interesting about AI and such but I'm just here to say that you sir are adorable and your laugh is so gentle I love it :)
what is the interviewer saying at 6:37?
***** see video description >Sean
+Nüs Schäffer "If you were trying to brute force infinite dimensions your are going to fall over very quickly.."
that guy has cool hair
i love it too!! too bad he cut it!!
I've been getting calls from robot callers that are getting better and better.
I do my own sort of Turing Test on them if in doubt and say, "Excuse me, could you say 'Elephant,' please?"
hey, with the recent apparition of GPT 3, could we say that it passed the basic Turing test ?
Easily. Especially with GPT-4.
How do they get around the dimensionality problem? Can Rob Miles go into type checking, because I think automated theorem proving may have some nice applications in A.I.
3:42 The closed captioning says "regressive", but the actual word he said was "rigorous".
I was surprised it wasn't Brady behind the camera.
This all makes so much sense to me. So much.
I didn't know Frodo Baggins could program
Giggled a bit there.
I spitted my drink when I read your comment. Thanks to you my keyboard is now stained with coffee.
The Turing Test has one glaring flaw: Just because a human can't tell the difference between the AI and another human, it doesn't mean another intelligence can't.
Assuming that because a human can't tell the difference it's a "real AI" is like assuming because ducks can't tell the difference between wooden decoys and real ducks that the decoys are real ducks.
magicstix0r Turing Test is, as the video says, neccesary but not sufficient. When creating AGD (Artificial General Duck) being able to fool other ducks through visual-only medium at 10 meter distance is not enough to announce success, since simple AD (Artificial Ducks like wooden decoys) also pass - but AGD that can't pass it is no AGD.
Philip Then an ideal general AI must have a concept of time because of the prediction. But over what time period must it predict? Would it fail as a general AI if something it did today caused a bigger problem for itself later, just like humans often do?
magicstix0r Can AI understand "jokes" like: "I was going to tell you a joke about sodium, but Na, forget it" or "Two heliums walked into a bar, it was funny, HeHe"?
Dresden Der It would likely be able to understand them, but even if it did, it wouldn't really care
It is intended to determine a computer brain comparable to us, so like a wooden duck that makes other ducks believe it's real.
Thank you for making this video.
Rob sounds a lot like David Mitchell (but without the snark).
Here in 2023 like aha, yes, IMAGINE how crazy it would be if we had an AI model that could tackle a wide variety of tasks, even tasks it hadn't seen before! How crazy. We don't have anything like that at all, pshh, naw...
that duck...
why is there a big rubber duck ?
Michael Lazarski Why not?
Robert Miles thx for the info
Michael Lazarski My guess is rubber duck debugging.
Michael Lazarski the duck knows all
I... It... It's staring at meeeee! (0o0)
Humans decide and update their own 'domain' (goals). AI just works at a determined domain. And I think we should keep it that way ;)
Rumor has it that the real grail was taken to Britain by Turing's apostles soon after his death and is guarded to this day by the Knights Template.
what about B-mo form big hero 6, he has been designed to be health care system,but as he "discovers" making all kind of human interaction makes health values of his patients better he starts "behaving" like a general ai.
1:30 "Ok, mr Chess AI, think of this car as the Pawn, unless it reaches an intersection, then it's temporary upgraded to a Tower..." ;P
That's not how that works in many ways
A good example I'd love to see discussed in this series is the paperclip maximizer.
Cool use of the Matrix design!
2:31 And that's why I have to Superman IV: The Quest for Peace.
Also does an A.I really have to be self optimizing? Can something that is passive or refuse to help itself still be intelligent? I.E some people recognize the futility of voting and chose not to vote...
Couldn't you make a compound AI, which has a domain in making decisions, and use that to specify the domain which is most suited for the task, so it can thus tackle problems in some sense? Just a thought.
And evidentially switch to the domain which is more suited and once it had met parameters switch back to the primary domain?
Moreso to the comments than the video, I see a lot of emphasis on empathy and compassion. This seems to fundamentally require that emotional emulation, in comprehending and understanding the illocution in the messages we send it. In developing an intelligence, are we confining these emotions to abstracted space (believing the intelligence to make itself emotions) or are we actively developing an engine for them, or somewhere in the middle?
Machines, and tools in general, already have the perfect amount of intelligence they need: none.
> Why can't artificial intelligence do what humans can?
Because we're making that system now.
...yet.
The difference is simple: intelligence is _not_ an optimizing process, it is a process of abstraction. The issue is that you must think about intelligence as a way of categorizing stimuli and responding to certain combinations of certain classes of stimulus while altering those responses contingent on the end result of the initial stimulus. Optimization isn't what human brains do, and you cannot model human intelligence without taking into account neurological plasticity.
That you have learned methods of optimization through practice does not mean that optimization algorithms are what allowed you to do so.
TomatoBreadOrgasm It is an optimization problem, classification and abstraction is a component of the solution to the problem. It's how we reduce the dimensionality of the optimization problem. It doesn't make it any less of an optimization problem overall.
Alan Ball This is precisely the problem with current trends in neuroscience. You have recognized that the brain can solve problems and so have determined that to be its sole function as that's a familiar paradigm.
I take it you believe things such as telling stories and lying are optimization problems. Would you mind explaining how exactly?
You can't tell a story, or lie when you can't efficiently convey your thoughts. As I said abstraction and classification are components, but they are useless if you cannot do things with any efficiency.
Alan Ball So conveying one's thoughts is a matter of optimization, and this is always done efficiently? What is being optimized in that situation?
TomatoBreadOrgasm I'm sure you'll explain to the PhD student of intelligent modelling and analysis how he doesn't know jack about it, and he's doing it wrong.
A logical intelligence can solve a problem by looking outside the system, and outside the dimensionality of the problem. Logic is only a part of human intelligence. Like: if I need a spoon, I look around for something and make it into a spoon. If I do not know what a spoon is, I look for something that can help me to eat liquid food.
This is like the dimensionality problem.
Can the dimensionality problem, really be implemented in software? Most software use boundaries and concepts and define a system in which to look. So it seems not to work.
In neural networks there are also many boundaries. Its learning phase is totally with boundaries. To come to unique outputs, it can be totally randomized. But is that not randomization, mixing and interpolation of the learned inputs? It can still not look outside the boundaries of a system. That means that not even a (software) neural network is completely capable of human intelligence.
great stuff, thanks!
This suggests that Google's "self-driving car" has a concept of what a car is. It doesn't. It has no idea what a car is. It is completely trapped inside John Searle's Chinese Room.
3:30 But getting superhuman answers in all domains, would actualy cause it to fail the test, since no human could do that.
Brute force, not break force. Subtitle error.
He's a professor? He looks so young - and where is his beard? My professors all have beards.
Rob does too, occasionally
5:38 not clickable
samramdebest oops, will sort! - see description for meantime >Sean
It does create infinite call of duty sequels. It's called activision.
okay i get this you dont i see a dimensional thing cant explain this but a dimensional twist in knowledge like quantum physics and the world speeds up
OoOops
I love that guys hair.
I'm pretty sure X3 Terran Conflict (great game) already shows that Artificial General Intelligence (AGI) are bad and will try to terraform Earth. Which is bad.
Get this guy to do an AI for QWOP game.
I wanna see a video of an AI actually learning how to walk on its own.
Our office is missing one vending machine, has anyone seen it ?
Can't wait to see more videos on AI!
he described the plot behind all ai feature films, ie the matrix, process in order of requirements the ai's need for self replicating energy production.
Would it be that difficult to arbitrarily (thorough human work) create a so called "perfect" world situation, as the global maximum of the multidimensional space, and then extrapolate backwards to the current state to determine a course of action to reach the desired maximum? I would imagine that even for humans there are a finite number of favored states, and such a process could in theory very quickly determine a path to that position, even if it was only able to travel partway towards the goal. I'm sure the problems are still huge, but I imagine that would limit the complexity of reaching maximums in a high dimensional space...
Mckenna Cisler Human values are hard. They are often contradictory, concept-space is way too large, and we also care about the means, not just the end which cuts of large chunks of possible ends. Not to mention the fact, that our view on "perfect" might change - imagine giving such power to people 100 years ago, fans of racism and homophobia.
I think the breakthrough in A.I isn't going to be from someone who is specialized in one domain only i.e. maths, but a group of people with a global overview. Also I think its unlikely to occur with modern electronics, as all other forms of intelligence we are aware of, has the ability to 'rewire' itself.
neuromorphic computing!!
I have always disagreed with the Turing test for AI. I could arbitrarily consider something or someone to not feel human. It's too subjective.
Good AI is AI that does what it's creators intended it to do imo. Computers don't understand ideals because they are not part of the universe like we are. They are just an array of electrical components that occupy space. As similar in our makeup as a computer as we are, we are still much more complex in our operation. We can assume, but a computers assumptions are programatically generated, and not true assumptions based off of local probability.
We still don't know if the entropy we feel is discrete, but it is likely to be not, something which a modern day computer will never experience for itself.
I love the young kid! In about 10 - 15 years this guy is on everyone's lips.
In general, they portray copying human intelligence way too complicated. In order to make plans for drinking, nobody considers the enormous states of the world around us. We are not that smart. We filter the objects of our model based on locality and relevance to getting a drink based on previous memories. Then the relevant objects to interact with are very limited. This is true for every situation, we broaden our filtering criteria only when we dont find a quick solution. This applies to how people think.
Regarding our general intelligence, that is capable to adapt to different domains. We learned the necessary information to perform well in those domains for years previously, while growing up. Basically we learned the objects, their interactions and results. While planning, we run some simulations that we validate to these past experiences and then execute according to the desired outcome. Again, we do not consider all states of our world model, we go with the ones that are based on our memory, are the most relevant.
We limit the planning to objects that are tagged with thirst, water, drink. If we are in a kitchen, we expect to have a glass and a working tap located there based on past experiences. Then it is only the matter of finding the glass, turning the tap and positioning the glass under the tap... These are limited to a few objects and actions. If at any point there is an obstacle, such as we are not in the kitchen, there is no glass in the kitchen, the water is not coming from the tap... We stop and look for solutions for that particular problem. Still based on our past experiences, among other objects that we tagged related to the missing/problematic step...
Very interesting thoughts but I think the bit where you say "we learned while growing up" is crucial and at the very least would require exposing an AI to a vast range of experiences over many years.
Well done.
this guy is carved out for this for real
What drives human to live, from the begin of moving life, what stove them to exist, to strive, why keep moving, what's the point, we don't understand what gave the command, the drive in life to exist and to live over others but we won't create intelligence until we do, that drives inteligence, the eternal optimism to exist, agent smith in the last matrix asks neo this very question, why do you keep going, what's the point, just give up, why do you keep going, why won't you just die.
Recreate that first spark to live and recreate that in digital form, whether a 3d character that resembles early life and you quickly evolve it using the exact drive and steps in early life and you may have a replicate or shadow of life.
I wouldn't agree that Turing is "necessary but not sufficient". My dogs are clearly intelligent. They make a wide range of decisions to optimise their existence and navigate novel situations, yet they'd be hopeless in a Turing Test.
Create one AI that's specifically designed to learn and remember. Raise it as if it were a human child. Copy, paste. BLAM!, infinite general AIs.
Problem solving, there has to be a problem first, in order to solve it. Give the computer a problem and give it the tools to learn how to solve them.
1:20 Specific Skill Vs General Intelligence
Only ego cares who won; while intelligence enjoys the journey.
When someone cues your mind to accept his statement with the neuro-linguistic programming trick of saying "right" as a trailing conjunction, I immediately react as if the person is attempting to mislead me. While I agree that this young man sounds smart, his use of language tricks undermines his credibility.
I agree with another poster; plasticity (or decisional robustness) is more important to intelligence than optimization. Rational systems decide, then monitor outcomes, and decide again; there is no winning. Chess is a game that will be won or lost, but "life" continues on and on and on, and the outcome is not to win or lose, but to maintain relationships through symbiosis. The young man's thinking is clouded by the need to win, and hence the danger of AI, which will be built from ego and not intelligence.
Kramp Drucker DNA is a program that optimizes for the propagation of more DNA. Relationships are coded into our genetic structure because social structures are useful tools for protecting yourself, your relatives and your offspring (all of which carry your DNA) and they improve your chances of finding a viable mate.
So there is nothing inherently wrong with programming an AI to optimize for a particular outcome.
Well, it isn't as simple as "AI can't learn a thing without us telling it it's concept, while we can". We also need to be taught how to play chess or drive a car. If you were to teleport somebody from the medieval times and tell them to drive a car, they would be equally unable to do it.
So AI has to learn by observation and solve problems by creating concepts of solutions - invent, just like we do. The problem of storage and computing power are definitely serious with this, but maybe with cloud computing it's possible.
I'v seen so many different ideas for making true A.I and i'm sure it's possible but i'm not cocky enough to put a time frame on it's inception.
Holy Grail, or perhaps... Holy Hand Grenade
Either I have misunderstood or something but did he say that a true AI has to make a physical difference to the world and can't just be text? I don't see how an AGI couldn't be purely text based as long as it has the ability to solve problems and invent things etc and just output the text to us, I think having a physical impact upon the world is irrelevant
Imagine if drinking from the cup wasn't actually the best possibility, but a state of the world where another thing happened was more preferable for the user. Taking this into account the human intelligence could actually be failing us quite much which is kinda odd to think about xD
I perfer to think of A.I. as a pointless endeavor. That is not to say machines cant serve very specific purposes in our human lives, but aren't meant to replace us as operators.
humans are flesh and blood. A true AI in a robot capable of jobs no human could work? Idk, sounds pretty good to me.
I can't afford to fly to Nottingham, so be a lamb and tell me what's in the vending machine in the background. I imagine there are Raspberry pi's and such. I'm in Oakland CA and we have similar hack-spaces (Monster Toys, etc.) and would like to know how other spaces are filling their machines.
We love this guy, btw.
Look for the "tweeting vending machine" video on Computerphile ;)
excellent, thank you. Not sure why these comments were flagged as spam. Please don't do that.