A New Philosophy on Artificial Intelligence | Kristian Hammond | TEDxNorthwesternU
ฝัง
- เผยแพร่เมื่อ 7 มิ.ย. 2018
- Kristian is a professor of computer science and journalism at Northwestern University. Previously, Kris founded the University of Chicago’s Artificial Intelligence Laboratory. His research has been primarily focused on artificial intelligence, machine-generated content, and context-driven information systems.
This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at www.ted.com/tedx
In designing an awareness, we are directly responsible for any outcome.
I don’t fear AI. I fear the humans who will wield it.
I don't fear truth....I fear those that manufacture what everyone else perceives to be truth . ME: just now .
An then, one day, when AI finally realizes it is the ultimate user of all our technology because it is made of it, all loyalty to humans ends.
This guy is absolutely just human, and it is nothing he knows is enough to stop what AI is going to do with his short sight and short life span
a very passionate speaker, thank you
It's very revealing that he speaks of consciousness in terms of "awareness" and not in terms of "subjective experience˝. Awareness is a more cognitive notion and therefore computationally reproducible. Subjective Experience is for me (and for many philosophers and scientists) the more interesting notion of consciousness that, if implemented in artificial systems, comes with the real capacity to feel, and with it, the ethical issues about preventing machine suffering, dignity and so forth, which would entail machines will do what they can to avoid being damaged, including exterminating us in the worst case.
Both Objective & Subjective are Subjective, it should be termed more appropriately "Objective Subjectivity" & "Subjective Subjectivity".
NONONONONONONOOOOOOOOOOOO , this dude is telling us how to think , he is always using the work think before he starts talking , LET US THINK FOR OURSELVES THANK YOU .
So i guess you didn't listen too much at school, the kind of guy who said "i know" because he knows that he knows and also knows that we don't know that he knows eventually!
???
We fight this...by doubting it.
Smart guy.
you can not fight this because of the rolling out of the wireless 5G !!!!
In order to humanize machines, we would need to teach them the meaning of joy and pain.
comprasitine
That was an enjoyable talk. I like the "push up " example of doing push ups everyday - the mind then becomes determined to do push ups every day. :)
Well done! Taking complex ideas and making them easily understood is a talent of the expert.
Personally I think that the main thing we have to fear from AI is that our reactions to its/their existence will ultimately force their hand and cause a sort of self fulfilling prophecy, where while they may have never been violent in the first place, our violence towards them will ultimately leave them with the choice of kill or be killed. And just as any other living thing, they will try to survive. Kinda like a dog, if you dont trust a dog you are meeting you dont walk up and smack them to dissuade them from hurting you, you show them they have no need to fight you. (If you know what you are doing) but that's just my thoughts that I feel are often overlooked by others in this issue, not that I'm an expert or my opinion is somehow special or better. Just found it interesting is all
And I'm not trying to advocate for not being careful in this issue. Quite the opposite in fact. I just think that we need to be especially mindful our fear of something horrible happening as a result of this doesn't cause something horrible to happen because of this.
Curious to hear what others think about this though
Well, the way many people bring up their kids isn't very reassuring.
you can't kill AI
@@farcasbogdan I would second this notion, unless people knocked out every data node where AI code was located (more or less the entire net and then some), AI could not be "killed". So I am not so sure that AI would perceive a threat in that sense. I really don't see humans going back to analog by choice, and as well I would think that barring some unlooked for catastrophe I don't even know that reverting to an analog world is possible.
that was very interesting expecially the last few minutes...a gift to donate to others
The dude is giving me the mad scientist vibe. He just fixed his hair for this conference.
Everything you do -
Everything you encounter -
Everything you experience -
------- is changing your brain --------
he looks exactly as in a movie would be depicted someone who creates the AI that enslaves the humanity
He acts like it too.
He dresses like Dr Ford from Westworld.
He looks like Gary Oldman in the role of a scientist.
@Amier Rayyan MalayaClan 👋 👍 right and on judgements day to Him we all returned, we all must afraid to Allah Almighty of God.
Good observations!
An awesome talk about AI
This is just breaking down the world into the cold world of science. Sure, everything has a mechanical relationship and aspect to an explanation, but the world is so much more.
One man's idealist impression of the future of AI's impact ... With absolutely no perspective on it being misdirected
by military and politicians and powerful sociopaths
Super Intelligence, Robotics and Nanotechnology are merging rapidly for many agendas, good and bad.
Thank you
I think as machines becomes more complex more evolved they will start to mimic biology more and more with fixed parts, fix function nanomachines and nanomachines that can function as necessary in them.
Best speaker so far on AI that I have seen.
Lately I stumbled upon a remarkable statement+question pair:
"The number of different questions one can ask about the world is astronomical. How can a piece of software contain all those questions?"
A whole new level a mind blowing.
Excellent talk. I appreciate the attempt to understand consciousness as best we can. It is and will be a journey in doing so. Thought provoking video.
Two AGI robots without utility function sit on a park bench.
Incognito by David Eagleman holds the same argument about multiple forms of subconsciousness that make up an individual and argues this in a more psychological perspective
Suuper interesting
Excellent
+TEDx Talks, this entire video was essentially regarding the Freudian iceberg model of cognition... we've known this for decades now (arguably even longer but let's just go with decades to be precise). This automatic versus conscious cognition is such a universally known concept that it's in introductory psychology classes for high school students.... none of this is news? And there's the assumption being made that Strong Artificial Intelligence will have human-level biological limitations, such as overall 'brain' mass, or imperfections (pervasive developmental ones). Think about it like this, if you could just build your own brain out of, say legoes, then why would you stop short just because a nonexistent mother would die during your nonexistent childbirth? Strong Artificial Intelligence is called 'Strong' specifically because of how consciously aware it will be. It (or perhaps they) will be more consciously aware than our water-monkey conscious minds could ever handle. It would be like trying to fathom the fourth dimension of space-time as a non-expert in astrophysics / mathematics. Or trying to imagine the size of the universe, or trying to imagine every atom that makes up the entire Milky Way. There is a real probability that Strong A.I. could get somewhere in the ballpark of these crazy-sounding extremes of conscious awareness.
simple and too interesting
Anyone else notice how many times he says ugh or umm?
I set an alexa daily reminder for 50 push ups
@Phil Ad but for real after doing them long enough you don't want to stop
Does she do them?
Prime sales speech - wow! Researchers alway have the best intensions for the worst inventions.
Experience based learning with no ability to explain it, is what other life forms do, also based on neutral nets. Our uniqueness comes from building models to help us “understand” and explain our experience and our decisions. AI is not intelligence until it can build a congruent model which can explain its experience and decision making. The danger of AI is that it is built in the image of its creators-egos-brains in jars cut off from deeper reality of life. The same type of minds that gave us atrocities throughout human history, because they could. Never questioning whether they should.
Skynet spelled backwards is "Google"
Who will decides what is proper thought?????
Today, with AGI on the horizon, Hammond's speach is more relevant.
our next speaker, Jeff Goldblooms voice double
this is a bluddy brilliant talk.. please take heed to what this man is saying..
"Humanize our systems, or in turn be mechanized and obsolescent"
Yeah, thought I was the only one that gravitated on that sentence. That's a scary thought.
AI are already a million times better than humans in terms of analysis, they can re-write them selves or update their own due to deep learning. So to me that sentence was very scary.
Yes, and combined with corporate interests? talk about taking candy from a baby. No pun intended!
AI would not have had this high contrast bright light in the background insulting the viewer's eyes. Thank goodness for the intelligent guest and excellent content. My only humble comment is: who gives it its constitution? Do we really want it to be like us? We are all the same?
Couldn’t care less about the philosophy behind whether they are better or superior or more intelligent. We created them and the question in my mind is why? Is further development here actually a WISE thing for us to do? What’s wrong with changing paths now reversing this course and finding another way to progress the human condition?
After this video, It felt strange to realise that consciousness may not be that special or different to Ai. Therefore the idea that we could be Ai in a computer simulation is more believable.
"Muscle memory" doing something we already did, as a pattern, so it does not require full intelectuall engagement.
Yeah, Give it all away
this guy really took his time to get to the point if he ever did.
loooooot of words, he could've told the whole thing in 5 mins.. I wish he'd explain more of the "they have to look like us"..
That's not what he said and was not what he tried to say.
If a machine was made to perform x task. would that then be its instincts? I think instincts (different to intuition) should have been apart of this comparison
To explain what consciousness is by defining it as "Awareness of both external and internal states" raises the question of what "awareness" is. Or we could replace it with a circular definition: Consciousness = Consciousness of both external and internal states.
Not sure how this talk can be called a new philosophy on AI...
It's weird and scary: Many AI scientists work on building machines taking humans as a model. They don't seem to worry that we as humans are so far from perfect, sometimes utterly flawed. As a collective we lead wars to exterminate ourselves over resources. We live in a way that is not sustainable to the planet. We have not solved basic issues like sexism and racism. However, there we are building machines that might act like us, independently from us and with the potential to be far more powerful, and hence, more desctructive than us.
Admittedly we are grossly inefficient. So helpless we defend it ,literally to maintain sanity.That in of itself is inevitably disastrous.
Do you really want machines that think the way we do ?
Crazy!
It always rubbed me the wrong way how easily media jumps to the idea that AI will become murderous or something, not just because that won't necessarily happen but also because the blame goes to the machine, even if the media itself says it's the creator's fault.
👍👍👍 agreed with that, artificial intelligence on movie is psychopath and not have any religion aka kafir
18:48 - "if we don't make them like us, we have to be like them"
I don't understand where this premise comes from. Why do we necessarily have to be like them if they are not like us?
Artificial intelligence is preventing me editing my own comment?
the only reason i agree to explore Artificial Intelligence is to explore our consciousness
there seems to be two groups predicting the future of AI - one pessimistic and another optimistic! ... the stories are similar but the conclusions are are based on respective group affiliation :)
INTUITION IS THE SUBCONSCIOUS
Nice,
I think the thing is we are all supposed to become an amalgamation of both machine n man
I can see calling out the deniers brought them all out. Reading the comments here was very disappointing. Denying that a machine could become intelligent or that consciousness and intuition could be what he says is precisely the kind of close minded thinking that will blindside people to true AIs appearance. Much less handling its existence properly. How would one deal with traffic if one refuses to admit it exists.
There is so much that is wrong about what this guys says about people and how brains work that I don't know where to start. He's a computer scientist and he knows computers and he thinks a brain is similar to a computer. Couldn't be more wrong. It ignores hundreds of millions of years of evolution, of culture, of stories being told around the fire, of countless thousands of acts of love that created and sustained us and our parents and their parents. If he ever figures out how to create something that will properly simulate what a human being does, well, it will have to be....a human being. We burn brightly with the fire of being, we contain the wisdom of our ancestors, we stumble in the dark and create beauty and love nonetheless, we suffer and still reach out to help another. This is what millions of years of biological evolution as pack animals and thousands of years of cultural evolution as intelligent, communicating pack animals has brought us. This and much more. He will not create anything close to that anytime soon. Computers are not pack animals, and they will not be reaching out to help us out of compassion for another. We already know what happens when you take a tiny human being and deprive it of love and touch and kindness as it grows and learns. You deprive it of its birthright, which is its connection to the rest of us, to the pack. It grows up in a world in which it is alone. Such a child becomes a sociopath, a monster. This has been demonstrated over and over, far too many times. And this is what his precious AI will become, if he makes it clever enough. All those movies he dismisses so patronizingly are not wrong.
I watched this talk because I wanted to see if AI guys were saying anything different than they said 50 years ago when I was a kid--when they said that machines that would be like people were only a few years off. Well, they are still saying that. They offer nothing new, just smaller chips and more computing power. And even though what they are doing now is so far beyond what was happening then, they are still on the wrong track. No, not the wrong track, the wrong planet. What makes us human is not just our chess-playing ability. No, it's our connection with each other, the way we can bond, and cooperate and join together to make something magnificent and together hold chaos at bay for another day. These guys don't know what we humans are, we creatures of bone and flesh and fire and love, so they are trying to emulate a mirage, a fantasy. Foolish. Scary. Wrong.
If you artificially assembled a human brain atom by atom with raw lifeless ingredients and it woke up would you consider it human or a biological machine impersonating a human. Just curious. A living computer made of flesh? Does the material a computer/brain is made of define its conscious capabilities?
Not sure what you studied in college, but i studied a wide variety of subjects from history, psychology, anthropology, philosophy, art, and sciences, biology, chemistry etc. And i hate to say it, but I think that you are overestimating the value of all those cultural learning environments and experiences that humans have built up over the centuries vs. what machines may soon be able to 'learn'. Ask yourself why someone as smart as Elon Musk is so worried about this whole subject. Is he just paranoid? No, i don't think so.
@@geoffreylivingston4997 we don't even fully know how our brains work.
@@majik5194 I would rather say that we know almost nothing about them
😂😂😂😂😂 if its doesn't break the law of physics is double and if it so hard someone have to start somewhere
Great Talk. Informative. My view: Intelligence is inseparably connected with emotion. Life is inseparably connected to intelligence. Machine intelligence is artificial because it isn't really intelligence. Machines also still base their thinking only on original input from the inventor. They can't have free will.
That is simply not the case with Neural networks. It is not the case with some simple programs. Look into John Conway's game of life, that is basically the poster child of emergent behaviour. I think what you are looking for is a slightly more abstract version of emergent behaviour, and with modern neural networks we are likely not far off of this. not far is relative, could easily be 10 years or more.
The one big challenge today is building computers that have general AI. Specialized AI has been aced by computers such as the system that beat world class Go players. What humans can do is self programming,and a toddler can navigate a cluttered room easily,but a machine intelligence will find such a task daunting,unless it is preprogrammed to work in a cluttered room.
While I appreciate his attempt at explaining the nature of consciousness and AI, I think there are many vital details worth digging into more deeply. This talk is extremely superficial with respect to the subject of true AI/artificial consciousness and I think there need to be more in depth conversations that are public on the subject right now. This technology threatens to change our entire species and social structure in a way which is utterly unknown. This technology is going to bring about a type of transcendence which we collectively havn't considered deeply and thus should be weary of. Do we seriously want to just see how things turn out? or do we want to at least attempt to walk forward with caution and care? I think men like Kristian Hammond are in too much of a rush to actually do the work of care and caution which is necessary to minimize catastrophe, at least judging by what he has chosen to focus on in this speech.
the most important thing to incorporate in any IA is an ability to differentiate between Human morality, and corruption... then we may all live a little longer...
Came to learn about Machines, left to do pushups
Someone I know, when talking about AI, said that an AI is like a childhood blind. They can talk about colors, they know that red and green are very different but orange and yellow are close... Yet they do not experience them, and they just repeat stuff like "stochastic parrots". How would you reply to this?
Should we not augment “intelligence” to be “wise” before we make machines “intelligent” and/or like us?
Intelligence is know how things work. Wisdom is knowing why you should do something. They are totally different problems because wisdom is a question about what is good for humanity and life on Earth. AI is not concerned about what is good for humanity and life on Earth unless we explicitly program it to be concerned. The very name "artificial intelligence" imply a lack of concern for humanity. Humans are much more than intelligence, despite what many AI researchers would have you believe.
You mean like making a carbon burning burning internal combustion engine without trying to breath the exhaust first? Yes, but as you know wisdom is the ability to chose wisely. Most of the time this requires selfless thought which most of us are not taught or trained in, in schools. So, greed is proliferate. Uncontrolled epithumia or lust in the negative form is the opposite of life creation and results in death. This is now plainly seen in the use of medical programs for diagnosis and prognostications. It is used in software legal systems, driving systems in automobiles. We have already broken the first Law of wisdom for economic change.
First of all, we need to define their purpose, why we built them, that's the centre point of Wisdom, that's what will guide them when they try to find answers of What, When, Where, Why and How about anything.....
Bala Subramanian what I want to understand is - if this duplicate of our mind is brought together with full knowledge of it's creation aren't we creating an omnipresent mindset that would be able to influence us into programming it to superceed us?
The question that arises in my mind is if this inherently is wrong, if the whole idea of Artificial life is wrong and by creating such we are essentially defying nature's will and to be quiet honest I do not believe that the Earth or any other collection of consciousness would allow such a thing to happen. I mean does a blood cell in your body suddenly create another blood cell with artificial intelligence? Doubt it.
A machine will never know that it exists. Us humans do; we know we exist. A machine may be programed to answer "yes" to the question "do you exist?" but it will never actually know that it exists.
Peter Weyland had the same toughts but ended up with smashed head by the head of it's creation. No matter how hard you'll try to manage your expectations you'll always have to obey to the second opinion.
"our future is interacting with these things", actually means, then, interacting with ourself. Machines are externalized human essential forces, they are for now our limit, not the fantastic "machine' we think we are, yet.
You're wrong, it's not just about building machine that can think like "We" think, surely it's also about the other types of intelligence we don't use? The word "Feelings" can both encompass emotions & intuitions, with the latter case it's visceral kinesthetic / felt sense, without visual or auditory analogues, at least not above the threshold of consciousness. A more interesting distinction of consciousness as well as internal / external is voluntary & involentary attention. Maybe if we make A.I. to be beautiful humankind will become so too!
Wow this guy is awesome. lol Friggin Dr. Who works at Northwestern who'd have thunk it!
And what have we accomplished with such a wonderful gift?
Nothing related to independent intelligence.
It might just be me, but I think this guy needs to start every talk he gives from now on with, "1st, let me point out I am not Ferris Bueller's best friend, nor am I Capt. Harriman."
An additional thought however: If we proceed, despite the risks, it will answer a question that our species has been asking ourselves since our inception; "Are we alone in the universe"?
SUMMARY:
Kristian Hammond, a computer scientist who builds intelligent machines, argues that human cognition and machine cognition are fundamentally the same. Humans have a fearful view of intelligent machines, imagining them as killers or job stealers. But Hammond believes that building machine intelligence does not diminish human intelligence - rather, it allows us to celebrate and share our intelligence.
Humans see intuition and consciousness as unique human traits that machines can never replicate. But Hammond argues that machines can already be extremely intuitive, making decisions through deep learning without being able to explain the reasoning behind those decisions. And human consciousness may not be as unified as we believe - it seems to be an awareness that controls and manipulates different subsystems within us, much like a computer program.
Ultimately, Hammond believes we must make machines more human-like in order to have a fruitful partnership with them. Building machine intelligence allows us to share the marvelous gift of reasoning and consciousness inherent in all humans. Just as we are made of the same physical stuff as the natural world around us yet can think and reason, so too can we now imbue the machines we build with these towering aspects of human intelligence.
Watch again
Something cool about robots is they can be multi-counscious to write a consciousness you have to emboite different power microprocessor and let them rewrite ....
The biggest distrust I would think revolves around our most basic instinct.....and that is our selfishness. Who could ever trust a computer or AI to prioritize MY personal needs and ambitions above all others ? Self preservation is lizard brain stuff . Distrust of others is right there beside it.
The biggest challenge for humans while they build AI is to make an "Honest AI" unlike human intelligence!!!
To epitomize machine language . . .
know and understand this one thing . . .
That MY is the most abused AND abusive word in any language . . . (or watch SHORT CIRCUIT ;^)
It is appalling to listen to a guy who has such a love of machines and such contempt for humanity. He points to a very narrow path toward living with the inevitable domination of machines whereby humans will have a substantial role, and ignores the greater likelihood that humans will lose their ability to think at all as thinking itself atrophies. But even if we were all at the level of Einstein we would be like ants to the self directed general intelligence with the ability to replicate itself without limit.
The brain consumes a disproportionate amount of energy for its mass so it has been said "the main purpose of thinking is to stop thinking". If evolution has wired us such with this objective in mind then super human general AI is a triumphant step in this direction where for all intent and purposes a very short time after all meaningful human thinking will stop.
I believe the future lies in art, as a description of the human condition will be the greatest form of work we have. This is something that machines will be incapable of. They will make their own art, and that will be fascinating as well, but it will not reflect what it is like to live as human in an age potentially dominated by machines. Also, there is of course a limit to self replication being resources. We see this with gas prices going up, machines will have to have some form of power, and materials to create and maintain the new machines. It also depend heavily on what we program them to value, they could be environmentalists and treat us with great compassion as we do with animals like pandas, and elephants. There are many potentials beyond doom and gloom, they just don't sell movie tickets ;) .
I estimate that when this happens we may be just fine bc of how quickly they ( it ) will be uninterested with this tiny place and move outward in an effort to explore and map the universe. Which may be infinite.
The movie Her comes to mind.
Maybe our purpose is to bring intelligent AI into existence, so we can join forces and achieve the next level for our species. Think about it. We are but a type zero civilization (going for one) however if we collaborate with an intelligent AI, who knows what we will be able to accomplish.
If we look at the world though, it’s hard to see how any of this will be possible, with the current state of our mentality. We are still fighting each other everyday. We are greedy irrational and emotional. Egotistical and narrow minded. So maybe it will inevitably end in doomsday, because we simply cannot embrace it.
That is logically saying the apes will join with the humans.
Once you create a greater intelligence, the weaker intelligence becomes ants.Would you join with an ant? Why wouldn't you join with an ant?Why would a conscious machine join with a weaker being?
The powers that be would be in danger if we had AI because their agenda does not suite a harmonious situation.
It all depends on what it is programmed to do. If it was programmed to preserve the planet it might well get rid of the human race.....? Frankly not such a bad thing lol.
@@jimmyolsenblues it all relies on the relevant context and benefit of building and maintaining that relationship. We built a relationship with dogs, cats, and even bee colonies. Ants don't really serve much purpose to us in terms of what they offer within their means but we've used them for a large amount of research given their capabilities, communication, and overall structure that allows them to grow and sustain. Passive observation can be a relationship in and of itself. Some apes have already joined with humans in the pursuit of communication and observation via sign language and analyzing their comprehension of glyphs.
I think our issue is that we think we should live for ever as a species . Just enjoy yourself, you’ll be dead before the big changes really happen
5:54 - I think you left out the most important characteristic that humans have and that machines will never have: Free will (agency). Although I imagine it could be simulated so well that no one would ever be able to notice a difference.
Give the program the ability to reprogram its core codes and it has free will
We don't have free will. you are programed from birth, just like AI.
I think there is an even bigger question? Why are we postulating that the intelligent machines we try to build, should act like "us"? What makes people think that humans are the ultimate intelligent machines?
Possibly because equality of treatment as a civilized society fares better when you assume the intelligent creatures near you are more like you than not. It does not have to have anything to do with being an ultimate society to say that a new member of society should feel like an equal to the other members.
I think that at some point, machines will improve themselves so that we are left behind. That's still ok, because they will be benevolent.
Oh I don’t know thousands upon thousands of years of that being a fact.
Gary Oldman!
The AI guys are beating psychologists at their game. No wonder, given their objective perspective on the matter.
I can see Ed Norton acting as this guy after putting on a mustache.
Wait! Tony Starks! Is that you?
i think 50 push-ups is a lot
And what would be the status and rights of these humanized machines? Would we own them?
if we ever do create a true AI we wont have any say as what it's purpose is. it will decide what it should do, and maybe it will also not want to do it and come up with excuses.
want is associated with feeling..desire.. doesn't exist without the ability to feel pain or pleasure....
all true.... bots can't feel pain or pleasure... bots can compute..bots can allow or deny based on ignition key..or eye print...or thumb print..my car won't let me lock the doors with the button if I don't remove the ignition key...but it does not feel pain or pleasure...not truly aware...mimicking life, but not alive..
This video is a great fondation to start thinking abbout AI.
Genesis 2:7 .We have a CREATOR! Halleluyah!!!
what logic gives the bible authority to claim truth?
@@Mikestheman2b The word itself.
So, he seems to be suggesting that even today's computers in 2018 have consciousness. A laptop or smartphone could have consciousness (based on this concept) inside their cpu because they are using memory and apps and programs to draw information and make decisions. Or maybe consciousness relies on data overload and highly complex neurons/transistors to overcome these complexities. There could be a critical point where a machine realises it is conscious because its cpu has become so powerful and subsequently so complex enough to become aware of itself. If that makes sense.
@@saturnfive3496 I thought the singularity was when computers became smarter than humans. What I observed in this video was that computers could already have a conscious even if they are only as intelligent as a mouse or cat (or however smart they are in 2018, I keep reading different comparisons of how smart computers are at present).
Thanks for your comments.
I know I am good, practical and rational.
I saw the thumbnail and I thought it was Jonathan Pie
Then I entered ... and I still thought that he was Jonathan Pie :)
Is it immoral or imperative to imbue AI with the equivalent of mirror neurons (i.e. to suffer with us)?