Intellectual socialism. People shouldn't dumb themselves down for people to catch up. The back of the class should put in personal effort to excel,not rely on being hand held through concepts.That logic has people in a sysyphean revert to base knowledge.

The concept of layering is really brilliant. AI literally learns the pathways and then tells it's buddies the paths. It's like running a million trials in 10 seconds. Look at something like a war battle...now you simply feed in the adata, the ai runs a million different scenarios, and picks the top 10. Wars over in a few hours if that. Same for like your medical issue...feed in the data points and the AI looks for a match. Who needs some Doctor and a pile of silly tests when AI can tell you via YOUR PHONE the exact issue you have and the way out? this is literally a few weeks from being real I figure. Once the DBs are made its simply an issue of flipping on the AI. They you use an app to answer some questions...2-3 mins you have your diagnosis and path to recovery.

I agree. When chargpt came out, there is this reaction that it’s just a word predicator, it doesn’t understand. I almost took the same view. But this is wrong. While the foundation of gpt is word predication, the demonstrated ability goes far beyond that. It is almost as if we throw lots of books to a kid, just asking him to recite them all. Later on, we are surprised to find he can not only recite but also write new content, answer questions or even deduce conclusions. This is exactly how GPT worked. We can’t say since we never explicitly teach the kid on comprehension so he must not have understood 😂

@@fool9111z you nailed it Bro.

true, and no wonder we don't know what is really happening in the model? which makes it even more scary

@@viv1902 I don't think this is scary because we also don't know what is happening in the mind of anybody 🤣🤣

The following is my recreation of an experiment Hinton related in the interview above (GPT-4 two days ago) : Me: The rooms in my house are painted either white, yellow, or blue. In one year the yellow paint fades to white. Which rooms should I paint with what color so that all my rooms are white in two years? ChatGPT: If you want all your rooms to be white in two years, you should consider the following: Rooms that you paint white now will remain white for the next two years. Rooms that you paint yellow now will fade to white after one year, so they will be white in the second year. Rooms that you paint blue now will not turn white in two years, so you should avoid using blue paint if your goal is to have all white rooms in two years. Therefore, you can paint some rooms white and some rooms yellow, depending on your preference. If you want some variety in the first year, you can paint more rooms yellow, knowing they will fade to white by the second year. If you prefer to have more rooms white from the start, then paint more rooms white now. Remember, do not paint any rooms blue if you want all rooms to be white in two years. Me: Here is another wording: Some of the rooms in my house are painted white, some yellow, and some blue. The yellow paint I used will fade to white after a year. Which rooms should I paint with what color so that all my rooms are white in two years? ChatGPT: If you want all your rooms to be white in two years, here's what you should do: Rooms currently painted white: You don't need to do anything. They will remain white for the next two years. Rooms currently painted yellow: Since yellow paint fades to white after one year, these rooms will be white after one year and remain white in the second year. So, you don't need to repaint these rooms. Rooms currently painted blue: You should paint these rooms yellow. After one year, the yellow paint will fade to white, and they will remain white in the second year. So, in summary, leave the rooms that are currently white or yellow as they are, and paint the blue rooms yellow. This way, all your rooms will be white in two years. Me: Why not paint the blue rooms white? ChatGPT: You could certainly paint the blue rooms white directly if you prefer. However, if the blue paint is dark or vibrant, it might show through the white paint or alter its shade, especially if the white paint is not high-coverage or if fewer coats are applied. Painting the blue rooms yellow first could help in two ways: Yellow paint might provide a better base for the white paint to cover, especially if the blue is very dark or vibrant. Since the yellow paint will fade to white after one year, even if the blue slightly alters the shade of the yellow, it won't matter in the long run because it will eventually become white. But if you're confident that the white paint can cover the blue effectively, or if you're planning to apply multiple coats, painting the blue rooms white directly could work just as well. It ultimately depends on the specific paints you're using and your personal preference.

Ùnua 3ua3uTtfģiò

And look at how they're all rushing forward to admit how wrong they were. Ah, 'experts': Untouchable Since Foreverᵀᴹ.

@@dixonpinfold2582 Well, they are all admitting how wrong they were. Which is what an expert should do. Constantly update their views based on new evidence. What else should they do?

@@squamish4244 Admit that they don't deserve the credibility they award themselves, that's what. Try to be a little less self-serving, that's what. Stop patting themselves on the back for being the most wonderful creatures under the sun, that's what.

@@dixonpinfold2582Yeah, humility is not common between people in competitive industries, be it academia, the industry or politics. But it doesn't help either, once pandoras box is wide open.

@antigonemerlin The video editing was done by a beginner. A professional would've known some chosen transitions are dangerous for people with epilepsy. Parts of this video is awful for such viewers, and potentially VERY dangerous (depending on their sensitivity, it will cause an immediate seizure -- a dangerous brain fault).

And, he covers a bit more ground here thanks to good questions.

It's crazy that programmers don't even understand what they've created. The Frankenstein metaphor is obviously there, but so are others. Alexander Fleming famously discovered penicillin by accident, and Albert Hofmann discovered LSD by accident, too. One has changed our civilization vastly for the better, and the other *could have* by now, if we hadn't f*cked it up.

Someone tested gpt4 on competitive programming problems from topcoder iirc. Theyre ranked by difficulty. It was able to solve every hard problem from before 2021 and none of the easy problems from recent year. Real understanding is what humans have, any human who can solve the hard problems can solve the easy ones too cuz thats how understanding works. Gpt4 is good at what its trained on because its a prediction algorithm but a very useful and impressive one. Someone should train a language model on some programming book for beginners and see how it does on challenges it hasnt seen before.

​@@Nxnn132J😊 I

😅😅😅😅

In the recent 60 mins Hinton shows us a bot which understands why a couple is selling their new baby clothes...their own child was a still born. the AI using the data provided concluded as much and was totally correct.

While I am hugely grateful someone of Hinton’s experience and authority on the topic is speaking, I am disappointed at the degree to which he seems to be stifling a more honest reaction to…that which he calls his realization.

@@vincentcaudo-engelmann9057 not sure what you mean? He doesn't seem to be stifling anything - he's just expressing his point of view

@@JasonC-rp3ly I totally agree, I mean i see him as visibly restraining himself on the view he is expressing

@@vincentcaudo-engelmann9057 Ah, well that is the way with us Englishmen - understatement is the way we overstate.

​@@vincentcaudo-engelmann9057okp😊p❤

Me too, although I personally won't

It is exactly what I was thinking as well

Great filter

Sam Altman or Ilya S.

Invention cannot be stopped

he's 75 and totally engaged. Unlike some Politicians.

Don't assume that because a person is a real expert at one thing, however advanced, does not automatically make them an expert in other things. Researchers are often not the best people to decide things like law and policy or regulation. They are often the least qualified to decide these things.

I think control is the wrong word. Pretty sure that's why people say alignment. We didn't control Isaac Newton or Einstein, they had human values, they cared about human prosperity. If we tried to control them, they wouldn't have been valuable. I think its the same with a super AI, if we try to control it, will will either stifle its value or compel it to resist us.

It's an inevitably losing battle for us, but the big question is if we can live with a superintelligent AI and we will be okay and able to wisely benefit.

Imagine if what happened in the movie "Eagle Eye" happened to us with AI

I wonder why this isn't at the top of humanity's agenda. Not a news item anymore, or so it seems. That's worrying.

And you paid him?🕳️

It's just like how all of Geoff Hinton's lab is credited with the nascent boom of Deep Nets. If HInton had died, just like he mentioned Von Neumann, and Turing died young. This idea would probably have taken a lot longer. But these AI will not die...I mean that is just scary. In another interview, Hinton said, we are just a passing phase in Evolution for these immortal beings we created...and that is just IT folks, we are creating our own immortal offspring, and we will die out, maybe that was "our purpose" by our creators, or this random entropic experiment called Life.

If I gave up, I would be more miserable than I am now. I have to do what I can, in activism, public awareness, funding, etc., if it stands a chance of doing anything at all to move the needle on this problem. I have to live, or die trying.

... I want all of my nieces and nephews to be able to grow up.

Humm yes someone has seriously edited him. The picture keeps changing. 38:00 I don't have any trust that this is really him or what he said. Seems like a deep fake. Made by chat bots.

How is it different?@@BirdNerdJC

I think his warning about superintelligent AI definitely is not science fiction. Stephen Hawking said: "Success in creating effective AI could be the biggest event in the history of our civilization.' 'Our future is a race between the growing power of our technology and the wisdom with which we use it. Let’s make sure that wisdom wins.' He warned about the dangers of thinking machines becoming superintelligent: "It would take off on its own and re-design itself at an ever increasing rate. Humans, who are limited by slow biological evolution, couldn't compete and would be superseded.' And later on in his final book: 'We may face an intelligence explosion that ultimately results in machines whose intelligence supersedes ours by more than ours supersedes that of snails.' He argued that computers need to be trained to align with human goals, adding that not taking the risks associated with AI seriously could potentially be 'our worst mistake ever'.

based

The only thing standing it that way for most is pyramids of legal obstacles. Which will be summarily tossed aside once the public understand how much better it is without them. No more cornering the market on something and then charging people to use.

I guess the video editor wanted to achieve an overall balanced light distribution, so parts of the video are mirror-flipped.

5 years max for this. The hold-up is not the tech, its the entrenched interests scared of losing their power. Unfortunately for them once the public sees the benefits of using AI they are done no matter how much they fight. Go use bard as you would a normal search engine...but with the AI act as if it is human and see what it does for you. Tell it you don't understand something. Or you worry about this or that. Ask it for advice.

Some serious shade on philosophy in that last section. 😅😅😅

AI is changing the very meaning of what thinking is. What life is. It will literally render us all asking the same question? what do we do now? I think the answer is the same one you ask when on vacation...wanna try that? or see that? maybe or maybe I'll just take a nap. Go watch the videos about an AI tractor which zaps weeds. That sort of thing is now a reality. Products and services are about to go to zero folks. The mind from scarcity to abundance. You will WANT to be a minimalist as it will be the richest state.

It's "pixels" , in digital imaging the word pixel is a combination of pix (from "pictures", shortened to "pics") and el (for "element") Source: wikipedia :)

Very nicely said!

He is a for profiteer for Meta. He has a money incentive.

Both Chomsky and Searle assert that weak AI is not sentient nor will it ever be. Moreover strong AI, if it is achieved at all, is a long, long way off. ChatGPT is part of a massive PR campaign by big tech. Hinton is just another actor in that campaign - we know almost nothing about the brain as it relates to the mind. This we do know, the brain is not a computer.

I agree that this attachment has probably hindered LeCun's judgement on the existential risk. But I don't think it's accurate to just state simply that Yann LeCun ridicules Hinton. He may have made some comments a bit along those lines especially in regards to the existential threat mentioned by Hinton, but that doesn't mean he's been generally disrespectful. One thing you also have to realize is that LeCun is 12 years younger. So it was slightly easier for Hinton to make this shift since he was considering retiring from hard AI science anyway.

He's got back issues so he's often more comfortable standing than sitting.

The "rumor" about scientists unalived by a robot which escaped through a satellite, after downloading how to repair itself, is True, then I'd quit too!

Professor Hinton has a back problem, that's why he is standing. Besides this, standing desks aren't unusual