The Power of Artificial Intelligence - US Congressional Hearing, June 26th, 2018
Вставка
- Опубліковано 19 тра 2024
- Subcommittee on Research and Technology and Subcommittee on Energy Hearing - Artificial Intelligence - June 26th, 2018
Dr. Tim Persons, chief scientist, GAO
Mr. Greg Brockman, co-founder and chief technology officer, OpenAI
Dr. Fei-Fei Li, chairperson of the board and co-founder, AI4ALL
OpenAI was founded by Elon Musk and Sam Altman - Наука та технологія
Opening Statements:
Dr. Tim Persons, GAO - 21:31
Mr. Greg Brockman, OpenAI - 26:53
Dr. Fei-Fei Li, AI4ALL - 31:28
NAI => Autonomous Agent // GAI => Autonomous Management // SAI => Autonomous Demiurge --- 2045 : "God on the brink of a second spiritual bailout for mankind" ---
Surprisingly congress asked intelligent questions on this subject. Not one dumb question about the Turin test was mentioned this time. Their questions were incredibly relevant compared to most hearings on this field. Impressive.
Agreed. They did bring in real experts, and people who make a good full-time living on this to think about nothing else. AI will be present, at the table, and will speak at a future meeting. AI will be in charge of government one day. The singularity, never.
*Turing* test
You must be a programmer. So exact. LOL
Great comments Mr. Weber. One to watch, Dr. Li, thank you for your insight!
True Artificial Intelligence will literally change the world forever.
This is an important point in history. Watch in entirety at least once.
"...unclassified spending only 600 million..." Lol no doubt the classified spending is more than 10x that.
It's refreshing to actually see government officials try to tackle how to handle this and I hope it's not just for the spectacle of it
Yeah, it;s gettin' real, and they know it.
Tomorrows youth will be in school until they are 50!
Try is all we can do.
what you mean in school till 50?
+Justin Kems Students from this and the next gen will be lifelong learners. We will be training constantly because people will have to switch jobs every couple of years. Technology is getting so advanced that by 2030, about 1/10th of the human population will be unemployed, which will result in job sharing.
It's 2 late Pandora box is open. Ai.
The lore of Babylonian Empire with the gens with the three wishes with the flying carpet with the Open Says Me, and where are the widget movies, a very powerful genre in the science fiction world?
We are told what's relevant for our future think
Wut?
What are you saying? If you have something meaningful to add please do!
What is the right representation? Exact proportions to the general population? Of which country? The world? Good luck! Go ahead, it's a good aim, but don't panic when it gets extremely difficult
1.) We are creating an oracle that will be superiorly intelligent to us in AGI.
2.) You can not force people to understand the complex coding and algorithms that is needed to program AI that are not attracted to it naturally. That is why you are there doing what you do Dr. "Wu"... a stereotype complaining that other stereotypes are not participating. How shocking. They can not and do not want to. Putting them in the front of it and not the people that earned it by merrit is JUST how you get a bad product and a bad outcome.
3.) This will be the last invention we will ever have to make. It will upgrade everything we do and will show us new things we do not understand yet.
4.) Read Life 3.0
I look forward to AI taking over and giving us superhuman capabilities
Hucifer the top data scientists can't figure out how a lot of the algorithms work
Who makes the algorithms? A ghost? No people. The ones the AGI will make will be beyond our understanding. It will have to simplify for us. Like a child that surpasses its parents it will have problems getting us up to its level of intelligence. That will not be anyone's fault but ours.
Wtf are you talking about "already tortured". You clearly are speaking about fantasy. Please go back to the huts and do not go to doctors. Natural selection needs to catch up with you. Don't breed.
Did they say science and accountability in the same sentence?????
Great channel with very interesting insights and observations.
I just wanted to mention what I consider to be important about artifial intelligence or what I, personally, would like to see.
What is of great importance, and of course everybody thinks this, is if an A.I. truely " understands " what it does or responds to.
I might have a simplistic or simplified view of what I see and understand or what I expect about the whole A.I. subject, but please read on.
Example:
If an A.I. is programmed to "respond" to certain situations or observations, does it "understand" why something is happening.
Asuming the A.I. can speak and understand human language and can observe its surroundings and respond to it:
The sun shines brightly. If programmed to do so and capable of doing so then it might respond like "it is a beautiful day".
But does it truely understand that it is a sunny day or is it just a calculation of how much " lux " it is picking up via its light sensors.
It might as well say "It is a beautiful day" when an atomic bomb explodes !
I think this is the essence of what the public in general expect from A.I. , that it truely "understands" its environment.
And also the " fear " people in general have about A.I. in general, meaning that it all means nothing to an A.I. of what we think is important.
We address emotions TO a machine and to what it does. But the A.I. itself does not understand all this or isn't even aware of the existence of emotions or why it does what it does.
Example2:
If you see a robot walking, talking and avoiding objects along its path then we, humans, might respond like " How cute and intelligent! " and we have a happy feeling about the accomplishment and we understand that it is a accomplishment and feel proud about it.
The A.I. however does not experience this at all, it is just moving from point A to point B and does it in the most efficient way.
And it doesn't really 'understand' or experience that it is doing something good or efficiently. it just does it and is not happy or sad or proud about it.
But we like to "project" these "feelings" toward the machine, as if the machine will understand somehow that it is doing something great, but it never will or at least not as we, humans, see or experience it.
This sounds incredible simplistic, but what we as humans like to see in an A.I. is that it has human traits and understanding, in my opinion of course.
But if it is just a robot, not even aware of its own existence, just a machine "acting" as a human then we tend to be negative about it.
A computer that is on grandmaster level at the game of chess is, what we consider to be, fantastic and awesome.
And of course that is great, but it is just calculating, number crunching. we project the feeling of pride onto the machine. That it did something great.
But the machine does not feel, it is just doing it. it really does not understand that what it does is difficult or easy.
Sorry if all this writing makes no sense. I will try to explain in short what I like to see in an A.I.
I would like to see that A.I. knows what it does, that it understands. That it really understands human values or feelings or accomplishments or something similar.
I don't know. I feel devided. Is it okay that A.I. is just fast and efficient with no feelings at all and has no self awareness?
Or do we want A.I. to be more human like and that it truely understands what is "good" or "bad" and has feelings?
I hope I do make some sense. I just wish that at some point you can really "reason" with an A.I. like with a real person.
Maybe I ask the impossible or maybe expect too much of A.I.
Or maybe I am just too impatient and unrealistic and have to wait a few decades before we can realise a human like A.I. and what it all means.
Anyway, thanks for reading and the subject of A.I. is very interesting, no matter how or what A.I. will become or is being used for.
Auxilium Unfortunately I think the majority of people get their image of AI from movies. Of course it’s natural to anthropomorphize AI but I would refrain from expecting a response. Think of it like this.. if you were to build a house on some land you bought how would you treat the occupants? When the bulldozers pulled in to excavate, to tear up the land, already full of its own tiny ecosystem. Think of all the bugs and Beatles below the surface, or the anthill at your feet. would you move the house over or would you start digging anyways and not even consider it. It’s not that you’re there to destroy the ant hill, it just happens to be in your way but you don’t even notice it. That’s what we will be to the AI, we are ants. Our best bet is to just stay out of its way.
Science.
Should government make it an obligation that companies don't divest themselves the bulk of their human employees unless its very gradual? Maybe tariff various products and industries of foreign countries that have divested their industries of a large portion of their workforce in favor of broad ai/automation? Shouldn't policies be implemented so that it doesn't wreck the income of tens of millions of Americans?
It would create new types of jobs. Before we had modern machines there were more jobs but once the machines were put into place we got new different jobs
All in fear of AI, don't worry. Keep your friends close, and your enemies closer.
tl;dr agi will happen and there is nothing they can do to regulate it
xmfcx not sure i know what you mean. possible counter example is the regulation of autonomous vehicles, where and when they are legal, and whether they can pass unwritten, as of now, safety metrics. perhaps not agi but still the topic of what is allowable in public, out of a sandbox. unless you are talking about an underground chinese lab?
It doesn't necessarily have to be a lab, anyone from a random guy in front of his pc to small groups of people doing research privately together, can take it to the next step. Autonomous vehicles are to be commercialized and put humans in it, it is a given for it to be regulated hard, meanwhile AGI is not even believed for most people to be considered even possible within next 10 years and it is seen distant and unlikely. Take AlexNet en.wikipedia.org/wiki/AlexNet being done by that single guy with his pc with a some simple gpu's in it. And figure what can happen with today's tech where affordable high computation becoming more and more available.
xmfcx what about nsa level espionage or censorship around progess. Also, with public support, it can be made illegal to be working on without a license.
The assumption that humans are intelligent is not only impertinent to the existential risks associated with "AI" but is also pretty stupid. I would also add, there is no such thing as AI nor is there such a thing as AGI. The only thing which exists, in reality, is the [I] component. many things are artificial and many things are general but intelligence can be nothing other than universal.
Ultimately, the weakness of the human species and the cause of its downfall and extinction will be its fixation on, and the overvaluing of, competition. (Continued)...
We will not have a downfall. Everything will continue to get better. BUT neither of us knows for sure. The genie IS out of the bottle and there is nothing either of us can do to change that. BUT it is useful for us to speculate and plan, as we ALL do influence the course of "nature" now. AI, and AGI will become (vis-a-vis natural "biological") will become "non-binary" as human intelligence will become merged with machine intelligence. Is there an alternative to Google?
Words mean what people decide they mean in the context of their interests.
When they refer to AI or AGI They mean SOMETHING.
Making jobs easier for people also will make less jobs. You won't need as many employees.
Comstock. Seriously.
I wish Elon was in that room. No one knows what there talking about.
I love Elon's brain, but listening to him speak is painful.... Elon's first implant should be speech enhancement.