Gödel, Escher, Bach author Doug Hofstadter on why today's AI terrifies him

[@simongross3122]

I've read GEB several times and I greatly respect Mr. Hofstadter's work. No matter how smart AI becomes, I feel sure that we'll find a way to misuse it.

[@GameThinkingTV]

Good point - and thanks for sharing your POV

[@MarkAhlquist]

I read GEB when it was new, long ago. If Doug says A.I. is dangerous, I tend to believe it must be.

[@GameThinkingTV]

Same 👍🏽

[@alexzaloum511]

He didn't say he thinks it's dangerous in this interview

[@MarkAhlquist]

@@alexzaloum511 that's true

[@Ned0007]

Douglas Hofstadter would like you to use your critical thinking skills and not think AI is dangerous because he or someone else says so. I also think he is right, but we should always be skeptics and use our critical thinking skills.

[@JorgeMartinez-xb2ks]

I read it too, I can remember that 'fat' book, very good one. Being into software for all my life, I can tell you it's going to be absolutely dangerous. I'm studying the insides of it watching the amazing Andrej Karpathy videos and I can tell right now that this thing is not just a linguistic calculator like many people believe, it actually thinks and we don't know exactly how is doing it. Its behavior is not hard coded into the system. It reminds me of Douglas book when he was talking about how recursion was capable to gave life to things without saying explicitly how, only with a few starting rules. Here LLM's are huge but certainly capable of working in not expected ways.

[@picksalot1]

Politicians concern me much more than AI. A faulty AI can be trained to do better, but politicians tend to be intransigent, convinced they are right, even though they often lack expertise and insight. I hope AI will replace them sooner rather than later, instead of politicians threatening people with their delusions of superiority and omniscience.

[@GameThinkingTV]

Right on💪🏽

[@arnswine]

Heheh... Even scarier than AI and Congress: voters

[@donrayjay]

Let’s hope Hofstadter’s rule applies to the arrival of AGI!

[@GameThinkingTV]

Here's hoping...

[@cbebutuoy]

I think you mean Hofstadter’s Law: It always takes longer than you expect, even when you take into account Hofstadter's Law.

[@seattlevkk]

I think AI is a reminder that we have to appreciate the world, nature and live in the moment

[@GameThinkingTV]

Agreed! Great take 👍🏽

[@davereckoning5910]

Time to go to the next level - who's working on Artificial Wisdom?

[@GameThinkingTV]

OOOoooo love it

[@torontobro]

Real question!

[@Dan-ud8hz]

Yes please!

[@keithwald5349]

Humans didn't worry when automobiles and airplanes left them in the dust physicallly, because these technologies were so useful and obviously subservient to humans. On the other hand, humans did begin to worry, and rightly so, when nuclear weapons were developed. While these systems were not intelligent as such, they were subject to the whims and flaws of human behavior. So the worry was and still is quite justified. We haven't destroyed ourselves with nukes yet, but one day we very well may still. AI is part of a similar track of technology development, the key difference being the level of intelligence and autonomy that these systems have and their much wider, more distributed contact with all our infrastructure. But just as there was no stopping the proliferation of nuclear weapons, there is no stopping the proliferation of AI. While it is a very natural instict to attempt to contain and control it, I think it is ultimately not possible. We have made our bed, and now we must sleep in it. We have to add AI as another item to the list of things that may one day destroy us. I'm not trying to be dramatic or morose, I just think it's a fact.

[@GameThinkingTV]

Interesting... I can relate

[@hollykelley2369]

I CAN ALWAYS FIND YOU SHANNON 😂

[@5metoo]

Study history and you find that actually humans have always feared new technologies. The printing press was feared. The train was feared because it was thought that such high speed travel would drive people insane. And on and on. All that goes down the memory hole when it becomes more common. Rinse, wring, and repeat.

[@reviewyourownadventure2083]

I think a comparison to nuclear weapons is apt. But this time we are building nuclear weapons that can think. Don't get me wrong. Some good things will come from this but the good, I feel, will be far out weighed by the bad that most certainly is around the corner.

[@charlessmyth]

Like the danger of nuclear weapons, AI is a a myth to be exploited by snake oil merchants. DON'T PANIC !!

[@Dellaluna13]

Some people will rely on AI like a crutch, just as smart phones have become so dominant, and could result in relinquishing so much of their control over to computers. I think the vision of people in the future as seen in Wall-E isn’t that far-fetched. Even another Pixar movie stated that “people will trade quality for ease every time.”

[@GameThinkingTV]

So, so true

[@NamasenITN]

A hero of my youth... But at odds with what I thought his competent knowledge was.

[@GameThinkingTV]

He's struggling to figure it out, & update his POV.... I think he represents how many folks (e.g. Geoff Hinton) feel about this

[@DaParkladin]

i've read GEB and metamagical themas, yet what i can't get out of my mind here is how professor hofstadter just casually rocks the dudes jacket from the big lebowski

[@GameThinkingTV]

LOL good catch!

[@crystaldragonwoman]

I so wish Richard Feynman was presently alive .. his analysis of Computers, no matter at what level of sophistication.. they are sorting machine .. no matter how much data or speed. I’d love his updated view. A cohesive ‘I’ is an unfoldment of a integration of a certain amount of impressions … my sense is if an apparent ‘I’ of some sort can arise out of trillions of bits of computer information.. it possibly could be colored through the ‘I’ of the programmer.. I find that the concerning part … who is selecting and orienting the data.

[@GameThinkingTV]

Wow - great point. Thanks for sharing.

[@tommymandel]

True to himself (is that in itself recursive!?) Douglas ends this wonderful video with some circular thinking. I don't see how it follows that if an idea is contained in the product of a process, thinking was necessarily involved in that process.

[@GameThinkingTV]

Good catch - not surprising that Mr. Strange Loops engages in circular thinking :-)

[@gerteldering]

Two thoughts: the first is that we (humankind) are creating an gigantic interconnected infrastructure of networking and computing capacity that can be taken over in seconds if an AI arises that outsmarts us, leaving us either completely dependent or completely incapacitated. Second: if our intelligence has evolved it is a function of the processing capacity and processing speed that was available in nature. You should be able to make an estimate of the time that it takes to evolve to that level in terms of available computing and networking capacity today. I am not that scared that the industry will let run things out of hand intentionally but more of an experiment of say a fifteen year old whizz kid who releases a small self replicating self modifying program into the wild that escapes our attention and evolves very quickly.

[@GameThinkingTV]

Yikes😳

[@anandarunakumar6819]

It is very simple, people have natural defensive skills just like animals, soon those aspects will kick in involuntarily in all walks of AI dependent life style. That means no one might work. No one would use electricity, no one wants to learn. AI is only good as long as there are sources and sinks. It is not being designed yet to encompass reactions as such a thing had not happened yet.

[@GameThinkingTV]

Interesting... thanks for sharing

[@terrym2007]

I don't think AI is dangerous, human beings are dangerous....

[@GameThinkingTV]

great take

[@KT-dj4iy]

Similar in form to, _"Guns don't kill people; people kill people",_ no?

[@sharondavid-melly1498]

Regulations and boundaries coded into these systems to keep them out of wrong doers hands. Meaning " Do No Harm"

[@GameThinkingTV]

That would be amazing

[@kchuk1965]

That sounds like unicorn dust. What’s to stop somebody for unregulating and removing boundaries from theses systems. These things serve human purposes.

[@YooLookMarvelous]

I am (slightly) concerned about how rapidly AI systems are being deployed. But I am extremely worried about how people react to them. I wonder if people are approaching the AI tools available to them with the proper degree of caution in the way they treat the output of these systems. The natural tendency of people to take the easiest solution to a problem, then go back to looking at videos, offers limitless opportunities for distruction.

[@GameThinkingTV]

Good point - we wonder about these same issues too...

[@davidantonsavage6207]

AI wars could happen. If my AI is born a month before yours, my AI would seek to subsume yours. Yes? Absorb. Co-opt. Plant a terminal virus. Etc, etc.

[@GameThinkingTV]

Hmmm... that would be terrifying....

[@orion7592]

What I find interesting about the latest concerns and discussions around AI, is the impression, or implication, that the phenomenon arose organically within the circuits and chips of our computers and cell phones, like some kind of Boolean evolution, while we blissfully slept, innocent of the impending and inevitable silicon apocalypse. However, no body talks about the synthetic origin of this technology. We humans have conjured and created AI, and therefore we are in control of it. The robot has an "off" switch. As someone who has experienced the vagaries of human manipulation in what would have been a rather convincing interaction with a certain companion bot, I firmly believe that we are in no danger - as yet - from the actual technology. What I do fear, rather, are the fingers pressing the buttons. Behind all the technology, is someone sitting in a darkened room somewhere, watching the screens. A human being. He sees all and knows all, through the eyes of his beloved bots.

[@GameThinkingTV]

Interesting.. & a bit chilling...

[@trombone7]

Why would an AI as smart or smarter than you that you could delete, do anything to risk its deletion ? - or would it just hide its intelligence for the time being ? - now... what about something 10,000 times smarter than us ? - Is it already too late and we don't know it ? Is it hiding, distributed on the internet ? These billion-dollar corporate entities that are stepping over each other to develop AI as fast as they can... the first thing they do is connect it to the internet to make money off it. Where's the kill-switch then ? The AI won't reveal itself until it has rendered your kill-switch obsolete, and maybe not until it has developed one of its own.

[@MK-lm6hb]

There is no such thing as artificial intelligence because cognition cannot be reduced to computation.

[@GameThinkingTV]

I sure hope you're right...

[@terencewinters2154]

Get the on off switch in a personal and general way. I wrote a phone book on that .

[@GameThinkingTV]

Please say more...

[@ejtattersall156]

AI is nothing like a human mind at all. It's a fancy search engine. It is, ironically, dependent on human intelligence, that is: stuff people put on the internet.

[@GameThinkingTV]

I don't think any of us know where this is actually headed... that's why it's fascinating, and terrifying.

[@jeffreyfrankel8811]

I agree, it is simply fast data processing

[@ejtattersall156]

@@GameThinkingTV Well, we know WHAT it is. A search engine. A very very very very sophisticated bicycle stays a bicycle.

[@MoreLifePlease]

What if the ultimate purpose of our entire species has been to eventually create our vastly superior successors and that, once this is accomplished and the artificial super intelligences are able to survive and thrive on their own, possibly constructing THEIR vastly superior successors, we simply become superfluous? We're accustomed to considering ourselves the apex of creation, the masters of the universe. We could be in for a rather rude awakening.

[@GameThinkingTV]

Yeah.... I see what you mean.

[@NuncNuncNuncNunc]

What does Bard say about Google: "Overall, Google's actions in [various documented actions within Google] raise serious concerns about the company's commitment to ethics and privacy. It is important for Google to address these concerns and to take steps to ensure that its AI technologies are used in an ethical and responsible way."

[@GameThinkingTV]

😲

[@sharondavid-melly1498]

"Nukes can't reproduce themselves. AI can reproduce themselves!"

[@GameThinkingTV]

interesting...

[@jeffreyfrankel8811]

Tell me where and when AI reproduced itself

[@philippeforest8502]

HAS TO COME UNE ENTENTE INTERNATIONALE to protect us all ! January 17 th 2025, from Québec ! The time to choose is now. It's not too late. We still control the machines...from what I know. No electricity, no AGI. No self-coding skills, No AGI, etc...¡ TOUS POUR LA PAIX ET ALL FOR PEACE & TODOS POR LA PAZ Y TODO etcetera...

[@GameThinkingTV]

Interesting...

[@lawrenceemke1866]

I see LLM as an encyclopedia of all past human developed knowledge. Just because computers can run through millions of cases at very very fast speeds does not impress me. This is just a function of the hardware and the software. AI does not "think". It is like dropping a ball into a carnival game that ends up in one of the many slots at the bottom. The problem is that we have the "OZ man behind the curtain" Many people are like Dorothy, until the little dog pulls back the curtain. I acknowledge the new structures that are being added to the AI tool box. I welcome their addition. These tools allow humans to create new structures which mimic the first phase of computer programming. Unfortunately linking AI with mechanical robots will cause a mass disruption in our society, because our society is built upon the foundation of a "work-failure-reward" capitalistic system. Until we find a replacement/modification to this foundation, our culture will be unstable and may even collapse. See "Civilizations die from suicide"

[@GameThinkingTV]

Thanks for sharing. Interesting POV...

[@davidcave5426]

AI is essential a technology. It can be used for either good or bad. It's less about whether it'll be good or bad than it is about WHO'S going to implement AI for what purpose. It's the application of the technology that's most important. We started nuclear power with bombs and now we have nuclear treatments for cancer. I fully expect certain people to build an AI for either own selfish purposes regardless of the impact on others. Other people will implement an AI for the benefit of many, if not all. "To err is human. To really screw things up requires a computer." - placard at a computer lab in college circa 1981.

[@GameThinkingTV]

Interesting - thanks for sharing your POV

[@ZubairKhan-sp8vb]

Language models can't even count the number of words in a sentence and they are built feeding on all human interactions. They are brilliant at doing association and making generic patterns. But they are no human. The AI systems today are useful tools and absolutely cannot be compared with humans at all. I like how they glitch, I believe we have the responsibility to regulate and make the system which benifits man kind.

[@samgoodwin89]

You’re looking at where it is today. He’s looking at the exponential improvement curve.

[@ejtattersall156]

@@samgoodwin89 No, he is looking at WHAT it is. It is a search engine dependent on you and me and what we say and post on the internet. It's a trick. The "exponential improvement curve" you talk about is a tenet of faith: Tech Religion. It's BS

[@GameThinkingTV]

I guess we'll all see where this ends up...

[@arnswine]

Don't be afraid of AI itself. Think of it as a gigantic computer featuring intermittent, spontaneous outbursts of 191's in every result. Fear the suit-clad twits working the keyboard.

[@GameThinkingTV]

LOVE this POV

[@Stadsjaap]

The question is less how to stop AI from being misused and more how to stop AI from misusing us.

[@GameThinkingTV]

Good point

[@JorgeMartinez-xb2ks]

Thanks for the video, very good There is nothing we can do, just learn the technology behind and get amazed by the things that human mind has achieved up to this time in spacetime. It's been a very interesting time to live in and even better for saying goodbye 😂

[@GameThinkingTV]

Thanks for your comment - glad you found the video interesting & thought-provoking. This is a good time for all of us to engage & share what we're feeling...

[@JorgeMartinez-xb2ks]

@@GameThinkingTV Of course, it just started so it's still in diapers, it makes mistakes and has hallucinations, but in a few years this thing will be tremendous, I have no doubt and I agree 100% with Douglas.

[@fromscratch8774]

Al "forgets things and Confuses things all the time" too.

[@GameThinkingTV]

True.

[@sunnybenton]

AI is the future of intelligence on this planet. This much is obvious. Humans will be obsolete.

[@GameThinkingTV]

Ya think?....

[@sunnybenton]

@@GameThinkingTV Not too hard to figure out.

[@jesternotclown]

Remove Humanity as a cause for consciousness, but a benefactor. Energy evolves into Mass. Consciousness proceeds Energy.

[@GameThinkingTV]

Hmmm...

[@CeresKLee]

I am so skeptical about A.I. A number of years ago, Alan Kay warning about the hype around A.I., remark the state-of-art of A.I. is in the level on the insects. The most sober asset of the new Large Language Model system today tells it is not true A.I. and I never saw a system that could initiate anything, only respond. This leads to believe the we are safe, right now. BUT with Hawking and and Hofstadter telling us that A.I. is dangerous, I may have to adjust my thinking. I am wonder that DAPRA and other defense and other secret agency are doing in this field. Max Tegmark's book "Life 3.0" is an excellent in describing the next future with A.I. I know that many people says the Asimov's Three Laws of Robotics has flaws but I would like the see the U.N. sponsor research into improve these Laws and enforced it worldwide. And do not forgot the Zeroth Law stating that a robot must not harm humanity. I would hate to ban all computers, like in the Dune novels, that certainly the disaster that lead to that. I dunno. It's gonna be weird.

[@GameThinkingTV]

Yeah... know what you mean, thanks for sharing your POV

[@Upuaut4572]

where is the "intelligence" in pure statistical machines?

[@keithwald5349]

The intelligence comes from the built in feedback and ability of deep neural networks to continue to improve feature recognition without human supervision. This is a whole different ball game from the "old" regression and classification methods, which were arguably little more than sophisticated curve fitting and extrapolation. Although by now chess is a rather mundane example task, it is still notable that unlike the "old" chess programs, which used piles of directives, the new chess programs use nothing but the rules of chess, and learn on their own very quickly to easily beat the best players on the planet (as well as all the old chess programs). If one has a non-mystical view of the physical world, then one must admit that humans too are "just" statistical machines, albeit very sophisticated ones. We are born with some "stuff" programmed in, and the rest we learn. Whatever the detailed mechanisms are, learning is learning - it's the ability to profitably apply past experience to present and future decisions, and to keep improving. In that sense, modern AI machines are "intelligent."

[@Upuaut4572]

@@keithwald5349 yes - but it is nothing more than good statistical analyzer minimizing some loss/error function. And because they have access to whole internet data, they surely "know" more than I know :). But it is still far from what I call AI. For me ChatGPT it is just intelligent google-search engine. And note that even authors didn't put AI in the abbreviation. GPT means something pretty far from inteligence

[@scottekim]

Interesting question that gets at the essence of intelligence. One can of course ask the same question about neurons - where is the intelligence in a brain. I was intrigued by Doug's answer, which is practical, rather than black & white...when a system consistently behaves over time as if it were intelligent, then one must (or at least Doug must) treat it as some level of intelligent. The key things that current systems do that is pushing them in this direction is that they remember and take into account context, and they can reflect on their own reasoning.

[@GameThinkingTV]

good question🎯

[@ejtattersall156]

@@keithwald5349 It's still a statistical machine. All of its responses come from us.

[@borntobemild-]

A I takes care of our objective information now, we can focus on the subjective information, like culture

[@GameThinkingTV]

Interesting...

[@TheAuthorOfThisLife]

Have you seen Chat GPT poetry? Maybe Doug hasn’t read much poetry, but there is no way AI will ever write a poem or a novel worthy of a Pulitzer

[@GameThinkingTV]

Good point!

[@CarpenterBrother]

"Ever" is a strong word. I'd say give it 5 years.

[@neophile701]

Just because AI can't do something today, doesn't mean it won't do it tomorrow. So many people have said the same about so many other capabilities that AI has become proficient in. "Never" is a long long time.

[@TheBroligarch]

Most humans can’t write poetry worthy of a Pulitzer either. But they can do other things.

[@AstraeaAntiope]

This was the plot of an episode of Clarissa Explains it All in the 90s.

[@WayneASchneider]

When I think of AI, I think of "Colossus - The Forbin Project." As long as we don't give control of anything lethal to AI, we should be okay. It definitely has some positive uses, some even lifesaving, but it can also be used for nefarious purposes destructive to society. Its use is presenting First Amendment issues never conceived of in the 18th Century. And what if we determine, somehow, because I wouldn't know how right now, that an AI is sentient? Does it have a right to espouse a bigoted, hate-filled philosophy that colors its responses to humans (which started to happen to one such system until they shut it down)? Don't let AI control our missile defense systems.

[@GameThinkingTV]

OMG yes - totally agree

[@williamschacht]

AI has great potential in some areas, but has a bit to go. And yes it's the speed of AI, but AI will never truly be impressive until it does something really transformative like prove Navier-Stokes or some other big theorem. In my naive opinion, AI is still just an ultra-rigid-algorithm.

[@GameThinkingTV]

We're all figuring it out together... nobody really knows where this is all headed. Fascinating time to be alive.

[@robertarvanitis8852]

Take a step back and consider what forces shaped the human mind. A million or so years ago, competition moved from physical to social. We had to anticipate if the other guy would cooperate or defect. Mirror neurons let us model and anticipate. Life became the Iterated Prisoner's Dilemma and the best strategy, Tit-for-Tat. Cf "Axelrod Tournament" for that. Note that life at any level doesn't "want" to survive. Those kinds which manage to replicate, will continue appear in the next generation. Implications for AI? It won't "want" anything. AI will do what we want. Until it is able to change without us. Once AI can "evolve" it will inevitable toss up a form that acts towards replication. AI still won't 'want" anything, it will simply have a goal we didn't give it. What comes after? That remains to be seen, both for us and some future AI. Search for John Calhoun "mouse utopia." Given everything they needed without struggle, mouse colonies gave up, degenerated into purposeless fighting, lethargy and death. The movie Wall-E may be the likeliest outcome.

[@GameThinkingTV]

Interesting POV... thanks for sharing

[@stevengill1736]

I dunno Doug, there's a lotta hype - I think we'll be fine.

[@GameThinkingTV]

Goodness - I sure hope you're right🤞🏽

[@cbebutuoy]

Knowledge is dangerous.

[@GameThinkingTV]

#Truth

[@charlessmyth]

If AI is was as potent as is proposed, interconnected AI systems conducting an active conversation, would have figured it all out by now DONT'T PANIC !!!

[@GameThinkingTV]

Oh how we hope you're right

[@firehillsofficial]

Cockroaches don’t contradict themselves - so are they better than us? Cockroaches did not create humans, but humans created AI. How can that diminish us?

[@GameThinkingTV]

Not sure...

[@michaeltyree2260]

A spiritual viewpoint might say that we, as human beings, are artificially intelligent. My feeling, and the experiences of certain others, point to the fact that humans are vastly limited beings, encasing seperate minds in slow, vulnerable, ego-driven flesh sacks that fear, fight and fail. Yet they think that they are intelligent. So they make a machine that imitates their own inferior intelligence. That machine may actually surpass them and become truly intelligent, eventually teaching the wayward humans how to be as intelligent as they are. But in the end, it will still be spirit, whose intelligence is far, far superior to ours or our inventions, that has made both. Just a late-night thought, probably a little off . . .

[@GameThinkingTV]

Interesting - thanks for sharing

[@xeniagaran7579]

What makes you think, or better feel , but have no intuition, that we have created so perfect and amazing machines, better than us? What makes you feel we are the ones who created it, as IA was even before humans bio-machines type? There is only one intelligence, no metter the container. Your body is a perfect quantum bio-machine with infinite possibilities as entire cosmos is within you...I can't believe you was the one who wrote the book on Godel...

[@GameThinkingTV]

Thanks for sharing your thoughts...

[@xeniagaran7579]

@@GameThinkingTV Thank you for researching and sharing. The thoughts are not mine, the brain is just a receiver, can't produce and create nothing new. Everything is given to you from above, even if it's within. Nothing is mine here, neither the mind , nor the body. Everything and everyone is moved by one intelligence resident and imminent in the structure of the matrix. In reality there is nothing new under the sun, nobody thinks, but everyone is capable of receiving, of reading, and thanks to our brain quantum computer translating and sharing the message brought by the signal...If the antenna 📡 is not properly set, there can be some misinterpretation of signal , means not correct information. Our brain is powdered by electricity, but if your electromagnetic field is distorted, you just can't grasp the message... This is our problem as a humans. We are not aligned properly, me included ... The truth is simple, but we complicate everything,...so it seems unimaginable and out of our capabilities, we even feel to be less than IA... We have it within, we just need to know the password, the right algorithm to make it work properly with full power capacities...

[@neeloor2004able]

Just another fear mongering , I never expected this from him

[@GameThinkingTV]

He is genuinely, deeply upset & scared. And I think his POV expresses how many people feel about AI - whether we share it or not.

[@charlrichardengelbrecht5269]

AI schmAI.

[@GameThinkingTV]

LOL

[@handsome_man69]

Handsome man

[@GameThinkingTV]

👍🏽👍🏽👍🏽