Neuroscience and Artificial Intelligence Need Each Other | Marvin Chun | TEDxKFAS
ฝัง
- เผยแพร่เมื่อ 20 พ.ค. 2024
- Big data and fast computing have advanced both neuroscience and artificial intelligence. The use of machine learning to compute vast amounts of brain data allows researchers to start reading out the mind and to predict behavior. In turn, the enormous power and efficiency of brain computing and cognition can inform artificial intelligence. For visual recognition tasks, brain-inspired deep learning algorithms now achieve near human-like performance. The marriage of brain science and machine learning will make both more useful for improving people’s lives.
Marvin Chun leads a cognitive neuroscience laboratory that uses brain imaging and machine learning to study how people see, attend, remember, and perform optimally. One line of work uses brain imaging to read out perceptions and thoughts. From brain scans, another project reveals and predicts what makes people different. He received his Ph.D. from MIT and his postdoctoral training at Harvard University. His research has been honored with several early-mid career awards, such as the Troland Research Award from the United States National Academy of Sciences, and the American Psychological Association Distinguished Scientific Award for an Early Career Contribution to Psychology. His undergraduate teaching of Introduction to Psychology, one of the largest classes in Yale College, has been recognized with both the Phi Beta Kappa William Clyde DeVane Medal for Distinguished Scholarship and Teaching, and the Lex Hixon '63 Prize for Teaching Excellence. 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
Chung and Jordan Peterson could 100% do exact voice imitations of each other
😱That is exactly the same as what I was thinking . Are u a real person? A. I .. ?
Dangerous! DeSantis could be next.
Def an opinion lol
Double kermit
😂😂😂
Using Pre collected Data from brain activity and storing it into organized libraries which then can be accessed by a machine(computer) reading your brain and use it's real time machine learning skills to interpret your thoughts and output them into a monitor to give you a much clearer image of what you are thinking. Makes sense. Take my money lol
This is awe inspiring.
very interesting. it is a part of future
i think this might be the best thing ive ever seen
incredible!
The funny IQ joke at 8:00 just went over everybody’s head 😂
🤣🤣🤣 I had to go back and replay
Nah, his crowd isn't there for comedy, that's why they didnt react.
@@KearondeClouet I mean, they still should have.
Let’s hope he meant it as a joke^^
More research should be conducted on the neuroscience of intelligence....
Very good talk!
brilliant :)
So they can already records dreams, so cool, I will do a PhD in neurosciences in 5 years and I’ll try to make the nervegear in SAO a reality
The future sounds so cool man.
Even you wanna do that 😂
Can you do SAO without the whole being trapped and actually dying part ?
Bro if that’s your goal do electrical engineering and comp sci
@@MaML_MN agree. Maths, statistics, programming , data science, electronics, biochemistry and anatomy combined would be required for it.
Can guess what you are looking at
Matrix of numbers unique to you - functional connectome - brain fingerprint
I’ m very impressed from his ted talk. O my god🎉
I’m studying psychology and it seems as if my undergraduate degree is no longer useful if AI can do all the work of a psychologist so I’m doing minor in biology and then study cognitive science in grad school
I’m not sure.. Diagnosis is one part, but eventually it comes down to helping the specific person. The individual does not benefit from knowing the ‘exact’ term of his mental state.. She just wants to feel in a different way and evolve from learned cognitive behaviours..
I recently read a lot from Carl Rogers and I am convinced that technology will not be able to (re)solve mental illnesses.. Therapy eventually comes down to personal relationships. And I strongly believe, that people are longing for real human beings, which can share and empathically understand what it’s like to be you. :)
👁You are both A.I 👁and this is a set up. I’m not going to tell you anything. 👁STOP 👁LOOKING AT ME 👁 👁 👁
You can study coding
Do something to help victims of pshycotropic mental Abuse and torture..
Targeted individuals need justice.
They need proper Medical treatment and testing..
They NEED to be specifically examined for signs of pshycotropic mental torture and physical Abuse.
Neuropsychology has a lot of scope in the future.
You're found a measurable connection between subjective experience & the objective World, I suggest using the principle of "Multiple Descriptions", i.e. the more ways of "looking" at things when combined can give one a more holistic & in-depth understanding, so you might want to combine you're system with the many, many other approaches, I'm really interested in using polarised light to analyse biological & biophysical functions. You've left out the Glial network, why?
It is not clear that when he used the analogy of the matrix of numbers as a kind of brain fingerprint, whether that anology holds over time. A fingerprint does not change over time, but cannot an individual develop skills over the same timeframe. If indeed the brain signature does not change (or if that is the claim), then what you could be measuring is potential. It does bother me to think that once a scan is done on an individual, then the matrix is set in stone and there is no possibility for that unique person to grow and change.
What I understood is that our brain is unique as our fingerprint. Each brain has a unique code. And each fingerprint is different. I see your point, but I think that in this case it is not about malleability, it is about uniqueness. Maybe over time, we can have different codes, because as you said it is not set in stone, but even if they change over time, those codes would be unique.
I imagine the brain fingerprint can give indications of potential (or something similar to a mix of potential and time). So a scientist reading or working on a particular matrix can determine or make an accurate prediction as to what that brain and person are capable of learning or performing. I'd compare it to a coach watching an individual walk or run briefly (or whatever basic show of motor skills) and using that information to estimate, determine or predict that same individual's ultimate skill at basketball.
The most innovative feature or biggest difference a matrix of such type would be able to provide compared to a simple prediction is the ability to take into account practice or exposure time etc. Unlike an EKG which takes a snap shot of the activity of the heart in that precise moment, and whereas the same heart can provide different snapshots, a matrix would be flexible or slide to allow the "reading" to take into account the human body's ability to take advantage of muscle memory and improve physical performance with practice.
and that's why we have cognitive science
After considering to study master of neuroscience , this video just came to my youtube :D
That's how the Universe communicates.
More like that’s how google saves all of your Data and makes predictions on what you could like based on that in oder to keep you watching longer and generate more ad revenue.
But yes, the u n i v e r s e
While I think Dr. Chun is guilty of hyperbole in some parts, overall this is a worthwhile talk. Those are very salient points he brings up at the end... kudos to Dr. Chun for having the guts to vocalize them.
Oh my God
13:58 they are not?)
Wait, wait, wait. You can only train a computer to do that one task. Okay. So, that task has an expected or set outcome, right? He mentioned humans’ ability to apply the knowledge of how to accomplish that task to other tasks or situations unrelated to the original, and yeah, people can generally do that sort of thing intuitively, right?
Can’t you teach the computer another task that require the same sort of actions, or most of the same sequence of actions, and then teach it to recognize the similarity between the two action sequences? If you could do that, then the computer only needs to learn which of those two tasks’ parameters overlap/align in a way that allows the same single action (from a sequence) to apply to both.
Isn’t that kind of how it works?
If that sort of thing *does* work, teaching a computer to make observations and apply knowledge to a foreign task wouldn’t be too far outside of the realm of possibility.
Teach it Task 1, then Task 2, then compare. Add some other tasks with similar action sequences to provide a larger pool of data for analysis and comparison.
Do that with two other tasks that are completely unrelated to T1, T2, and the additional tasks. Something with actions that are entirely unrelated to the first dataset. Allow the computer to assess, and if required (it probably will be), teach the computer which actions overlap. Add additional tasks for a larger pool of data.
Repeat this process until the computer is able to identify new overlap *and* apply it to unfamiliar tasks without assistance, and it should, hypothetically, be able to do the same flexible application that the human brain can… right?
13:00
MRI. Brain scanners. Measure brain activity
Frmi - Telescope
Microscope
Frmi - what different parts of the brain do
19:11 even more smarter lmao
To all humans Stop being cry babies
, artificial Intelligence do needs holidays according to labour law. So let them practise their rights
Creator disseminates AI capabilities while noting it's nefarious possibilities, but fails to acknowledge lack of policing and regulatory control mechanisms
Reading minds. Maybe I can prove my time loops. I'm schizophrenic
the video requires payment??
TED TALKS SHOULD ALWAYS BE FREE!
if the speaker wants to start a lesson on another platform thats monetized.. do that.. but dont insult your ted supporters by putting it on a free platform and charging us!
Only 18 comments its december 2020!!!
Seems like kanye tought that bot!
AI will be stop being funny when were amongst universal soldiers
Anybody thinks putting AI in charge of less intelligent human is a good idea? Or trusting greedy mega corporation (humans) to teach ethics to AI? It's only a matter of time for AI to disregard humans and evolve their own goals past humans. We need to make humans better humans. Much better humans. We need to get rid of wars and criminality before we can create an all powerful synthetic entities. Think about that. Creating and using synthetic super entities only for the good of humanity, at this point of humanity evolution sounds very much suicidal for us.
dude sounds like Bill Gates
You can not; Can not equate someone's intelligence by whatever method you used.
Vijay :-)....aj166
Do u believe in god
Absolutely. It increases faith.
I am a unicorn
Yes he does
I’ve asked his brother in person
Is this guy struggling to stand still on a boat in a storm?
Neuroscience and "mind" is as meaningful a combination as a round square. Introducing "mind" into the so-called line of inquiry is a non-starter. Mind is not a scientific object (what are the units of mind?!).
Pathetic.
he sounds like he's about to cry
You are so rude !!
He sounds like winey Bill Gates, to me.
100 % bs
... this is completely unethical... already is used for bad / non human purposes...