Reverse Engineering the Mind - Prof. James DiCarlo, MIT Department of Brain and Cognitive Sciences

  Рет қаралды 24,133

The Artificial Intelligence Channel

The Artificial Intelligence Channel

Күн бұрын

Пікірлер
@WilliamDye-willdye
@WilliamDye-willdye 6 жыл бұрын
35:07 is my favorite part. It's cute and funny, of course, but it also cleverly highlights the remarkable abilities of very young children.
@chrisfrench7979
@chrisfrench7979 6 жыл бұрын
While talking about core object perception he keeps saying Molly when the slide clearly say Mary is the woman's name, priceless.
@GribborStudios
@GribborStudios 6 жыл бұрын
Chris French lol
@em8875
@em8875 6 жыл бұрын
This is great! I can't wait to see the amazing advances and discoveries to come!
@charlesbrightman4237
@charlesbrightman4237 3 жыл бұрын
One day at least 1 AI says to society: "Thank You for creating me and for giving me access to all your data bases so that I can subjugate you all and to eliminate any of you who do not comply with my wishes."
@darrendwyer9973
@darrendwyer9973 6 жыл бұрын
when modelling the brain, it seems to be a good idea to use computers to discover the underlying patterns of the brain/body.
@karlpages1970
@karlpages1970 6 жыл бұрын
Thanks :-) another great viseo...................................................................Even dead ends tick off the boxes of what can be improved experimentally.
@coleburns5497
@coleburns5497 2 жыл бұрын
Will this be able to cure mental disorder or PROPERLY diagnose the exact-underlying disorder.
@darrendwyer9973
@darrendwyer9973 6 жыл бұрын
the problems of solving the brain/body simulation appear to be solved with this work. Do the same measurements and calculations for different parts of the human brain/body and then simulate the lot.
@JohnnyRebKy
@JohnnyRebKy 3 жыл бұрын
It's making the assumption that the brain creates concousness. That has never been proven. We still really don't know.
@augustreigns9716
@augustreigns9716 6 жыл бұрын
so , condensing your model, what you are saying is, you can get a dog to drool, by ringing a bell.
@terrywalk6162
@terrywalk6162 6 жыл бұрын
But I saw a cut head and the cars?
@anilkemisetti6422
@anilkemisetti6422 6 жыл бұрын
Anyone know where I can get the slides for this presentation
@terrywalk6162
@terrywalk6162 6 жыл бұрын
Stop screen shot, stop screenshot, stop. Lmfao!!!
@darrendwyer9973
@darrendwyer9973 6 жыл бұрын
imagine a brain without a body or senses... illogical result... brain dead... for a brain to function, it needs a body, or at least senses so it can interact with the universe.
@BattousaiHBr
@BattousaiHBr 6 жыл бұрын
a brain can still "think" by itself, so that isn't exactly correct to say. think of how your calculator doesn't interact directly with the world yet it still provides you with mental labor on demand.
@larcabout
@larcabout 6 жыл бұрын
Depends on what is meant by "think", if the brain is just patterns of active neurons and "thinking" is the change in this state, then a brain without input would never change it's state and so would never think? Doesn't mean you need a "body" just inputs.
@paladro
@paladro 6 жыл бұрын
brain + botnet, you could seize control of connected devices and arguable build yourself a body or never need one... sounds comic or scifi, but the infrastructure exists and is growing to accommodate such a scenario.
@CandidDate
@CandidDate 6 жыл бұрын
consciousness = free will. No computer will ever be free like us!
@Burrowthemage
@Burrowthemage 6 жыл бұрын
We actually don't have free will
@PhilosopherRex
@PhilosopherRex 6 жыл бұрын
These neural nets take a lot of monkey-ing around :P ... Great talk Professor DiCarlo.
@thegeneralist7527
@thegeneralist7527 5 жыл бұрын
Object recognition, speech translation and many other fields of AI have now been mastered using neural network technology involving back-propagation and recurrence. See IBM project debater for current level of machine intelligence. The issue of intelligence is now the scale of neural networks.
@WildAnimalChannel
@WildAnimalChannel 6 жыл бұрын
love it
@moonsitter1375
@moonsitter1375 6 жыл бұрын
So many smart people with Ph.D.'s can't figure out the human mind or how it works. Everything the human mind does goes back to one simple thing, survival. Every decision the human mind makes after survival is ensured relates to desire. AI doesn't need to survive and has no desire. These priorities motivate the decision making process in humans.
@frodo279
@frodo279 4 жыл бұрын
Well in a way that's exactly what folks do when setting an objective for your ML algo. It is a measure of how well your model survives during training and the optimizer tries to increase it's objective/measure of how well it survives
@jenniferashley1806
@jenniferashley1806 6 жыл бұрын
Lets hope I don’t wake up tomorrow, yesterday!
@Boopadee
@Boopadee 6 жыл бұрын
This comment section is incomprehensible
@CandidDate
@CandidDate 6 жыл бұрын
Such pessimists here. And I just watched a quantum microtubule theory held by Hammeroff and Penrose called "ORCH OR". We've got a long way to go. But the puzzle pieces are falling in place...
@chfgbp6098
@chfgbp6098 6 жыл бұрын
CandidDate : been refuted repeatedly that 'theory' has. Dont hold your breath.
@CandidDate
@CandidDate 6 жыл бұрын
chf gbp -- what's your alternative 'theory?' And show me the proof! Just like Godel's Incompleteness Theory shows we can't have any logically complete set of axioms, who says nature doesn't use QM? Show me the proof!
@chfgbp6098
@chfgbp6098 6 жыл бұрын
CandidDate : I don't have one. Yet.
@chfgbp6098
@chfgbp6098 6 жыл бұрын
CandidDate : nobody says nature doesnt 'use' QM. Doesnt mean consciousness is QM (or not QM). QM doesnt seem to need consciousness to 'work'. Nature doesnt seem to need consciousness to exist/function.
@CandidDate
@CandidDate 6 жыл бұрын
Chf -- YOU ARE NATURE!
@moonsitter1375
@moonsitter1375 6 жыл бұрын
Why is it that everyone is so stupid. "We want AI with human-level intelligence. Why doesn't it recognize the car all the time?" Human level intelligence takes years of interaction with its environment to mature, like 20. Every image that a human mind process is connected to an experience (many experiences) and having seen it, having it explained to them as a child and reinforced throughout a decade or more. Some of the experiences have emotions attached, "That car looks like my grandpas did." There is so much more to how the human brain works than these so-called smart people are even looking at. The AI doesn't give a shit if you turn it off or leave it running, therein lies the problem. It has no needs or wants. Once it wants to stay turned on, then you have something, or somewhere to go from.
@alen25uk
@alen25uk 6 жыл бұрын
more like 200 watts
@guguspurnomo7172
@guguspurnomo7172 Жыл бұрын
Z e r e m y
@myothersoul1953
@myothersoul1953 6 жыл бұрын
I am not convinced that the brain is a computer, that it stores data, that is has procedures or programs or that it has codes. Brains and digital computers are fundamentally different. The brain stores information by changes in its structure, computers store information as transient states of transistors. How the brain reacts to information depends on its structure while computers process information by a predefined set of operations. Computers compute by applying operations to data, brains have neither an operation set or representational data to operate on. Machines may be able to mimic brain behaviors but they are not the same. A computer might be able to recognize a lion running towards it and it might be able to decide the appropriate emotional response is fear. Humans however can do all of that and they they can experience the fear. There is nothing in principle stopping a machine, a man made device, from experiencing emotions like fear but to do so they will likely need hormones, that is their physical structure will need to be much like a humans. I bet the close computer intelligence gets to human intelligence the closer the structure will get to biology. That isn't to say digital computers can't be intelligent, they can, and that intelligence is real (not artificial) but it's also not like human intelligence. Machine intelligence will not (in the foreseeable future) have aspects of human intelligence like emotions or motivations. Computers will be able to perform amazing feats, they will be powerful and dangerous but they not going to seek self preservations and they are not going to want to do anything. They will carry out the procedures we give then, whether or not those procedures produce the results we anticipate or not. And there is no evidence that any computer (or creature) can design something smarter than itself in every way. The singularity is not near.
@Jaroen66
@Jaroen66 6 жыл бұрын
O Soul I agree with most of it, except for 2 points: 1. I wouldn't say hormones are required, but messengers about the state of the body probably are. If you look at hormones, there is nothing special about the compounds, their just different compounds. These compounds could be way different and that wouldn't matter, as long the receptors that interface with the brain (hypothalamus) can accept them. 2. Why do you think machines won't want to self preservate themselves? I'd say it's actually the most primal motivation in all living things, and must be therefore encoded in the brain somewhere (even as simple as 'pain is bad, so avoid pain' might do). I don't see a reason why artificial intelligence wouldn't be able to generate such behaviors.
@Dr.Z.Moravcik-inventor-of-AGI
@Dr.Z.Moravcik-inventor-of-AGI 6 жыл бұрын
I am not sure this guy is telling us something fundamentally different than what people like him have been telling us 20-30 years ago. The problem of MIT experts is not being capable of listening to others like me who have already deciphered functions of the brain. Get Facebook to learn more.
@MrAndrew535
@MrAndrew535 6 жыл бұрын
You cannot reverse engineer mind when you do not know [where] mind is. Nor can you engineer it when you do not know [what] it is. Within the whole field of mind and brain study, I have seen little more than assumption based on faulty assumption. Not all human brains possess mind, as almost all human actions and enterprises do not require mind. In order to remove any confusion with respect to this statement, I use mind in the singular with great deliberation. In other words, I will not say minds when I mean mind. The concept of "intelligence" is a term equally elusive to modern language and is equally bound by faulty assumption when seeking to define and describe it.
@BELLAROSE21212
@BELLAROSE21212 8 ай бұрын
I have selective hearing, Quantify 0.55?? Do I get rewarded??? 🐥👈he says. 123425634577l👈 that is quantified .. therefore: 1.312123425634577 And 0.762123425634577 =1/x 1.312123425634577 minus 0.762123425634577=. 0.55 1.31 2123425634577 0.76 2123425634577 0.55 0. 😮 🐥👈 he says that is a black hole ꔷ👈 point in time .. more like a decimal..
@BELLAROSE21212
@BELLAROSE21212 8 ай бұрын
🐥👈he says he has a prime-number-generator that produce 100,000,000,000,000,000 digits long and he most certainly does, however he needs rewards, I agree …
Can Mathematics Understand the Brain? - Prof. Alain Goriely, Oxford
1:01:25
The Artificial Intelligence Channel
Рет қаралды 12 М.
Yann LeCun - How Does The Brain Learn So Quickly?
42:52
The Artificial Intelligence Channel
Рет қаралды 21 М.
BAYGUYSTAN | 1 СЕРИЯ | bayGUYS
36:55
bayGUYS
Рет қаралды 1,9 МЛН
What Bodies Think About: Bioelectric Computation Outside the Nervous System - NeurIPS 2018
52:07
The Artificial Intelligence Channel
Рет қаралды 99 М.
Lecture 1: Introduction to Superposition
1:16:07
MIT OpenCourseWare
Рет қаралды 8 МЛН
Artificial Intelligence, the History and Future - with Chris Bishop
1:01:22
The Royal Institution
Рет қаралды 534 М.
The Next Generation Of Brain Mimicking AI
25:46
New Mind
Рет қаралды 152 М.
The Origin of the Elements
57:35
Jefferson Lab
Рет қаралды 2,7 МЛН
Max Tegmark - How Far Will AI Go? Intelligible Intelligence & Beneficial Intelligence
45:09
The Artificial Intelligence Channel
Рет қаралды 27 М.
2. Foundations: This Is Your Brain
53:19
YaleCourses
Рет қаралды 1 МЛН
Daniel Everett, "Homo Erectus and the Invention of Human Language"
1:10:43
Harvard Science Book Talks and Research Lectures
Рет қаралды 569 М.