Tap to unmute
Gödel's theorem debunks the most important AI myth. AI will not be conscious | Roger Penrose (Nobel)
ฝัง
- เผยแพร่เมื่อ 17 มี.ค. 2025
- Don't forget to subscribe to our channel and turn on notifications so you won't miss any of our future episodes ► / @thisisworldofficial
![Carabao Tawandang x NR กับศึกนัดชิงคาราบาว คัพ [ให้คุณดูบอลเป็นเพื่อนเรา]](http://i.ytimg.com/vi/2hAhPeUwHFY/mqdefault.jpg)
![[LIVE] : ONE ลุมพินี 100 | คู่เอก "เมืองไทย vs อิบรากิม"](http://i.ytimg.com/vi/m6ztjKGzN9w/mqdefault.jpg)
The interviewer is failing to understand Professor Penrose as he answers one question and interrupts him to ask another question that is irrelevant to the point being made. The interviewer is out of his depths
What a very polite understatement.
Yes, I agree. he is very irritating, trying to outsmart one of the smartest people on the planet today. He should let Prof.Penrose finish his train of thought
And his english pronounciation is annoying, for being an interviewer, at least you have to be clear enough
People like Roger have to deal with this all the time. I remember an old TV interview with a forced religious discussion element were he was constantly interrupted by an enthusiastic evangelist hellbent on talking about religion vs science who also happened to be on the same morning television (not a coincidence by the show runners, no doubt thought any academic could substitute for this role) effectively ruining the opportunity to listen to one of the greatest thinkers of all time
are there people like Roger btw lol?
The interviewer is doing a fantastic job of channeling Ali G without even realizing it. Is the interviewer conscious? I don't know
Back-fitting patterns and spewing out 'results' simply based on the fact that those patterns existed in the past (and arriving at those results through brute computational force) is definitely not intelligence unless the entity knows why the patterns existed in the first place. And like Penrose repeatedly tries to make the point that true understanding can only happen if the entity is conscious. The current 'AI' is just a bit more sophisticated search prompt algorithm - nothing more. Ideas, thoughts 'occur' to humans, a lot of times out of the blue. That can never happen with AI - it can only draw from the trained data - nothing outside the data can 'occur' to it. Godel's unprovable truths 'occur' to humans naturally and since those truths do not have a computational basis, AI can never get to those truths (unless explicitly fed those truths), never mind 'understand' those truths.
😂
@@najeeves8171 Explain that to the collegiate and trade education system... Thinking is not a graduating requirement. We need universities again (and not in name only).
He sounds buzzword/algo driven. Perhaps incentivized by the traction of “AI” and “quantum” in building audience$ with superficial treatments.
But he’s got a fine hairdo.
Man, that interviewer... Dude, CLEARLY you have nothing intelligent to offer in this discussion. Let Roger Penrose speak so we can all learn something.
No. An interviewer has to probe the subject. He did a really good job. One has to defend a scientific thesis, and Penrose explained his thesis completely.
I agree. He's interrupting, and to anyone who has paid attention to Penrose speaking before, it's clear he's missing the point he was about to make. He's also making the expression of someone who isn't quite understanding or disagreeing and is eager to interrupt 2:22. I had to stop viewing, and was happy to find this comment immediately as I scrolled down.
@@WalidDamounyProbing the subject vs. asking irrelevent and/or stupid questions. There is a difference.
That said, he improved later in the discussion.
@ That is called adversarial journalism. That is expected in both journalism and academia as well. This approach of asking questions is not something new.
@@toby9999 totally agree. Interestingly, I think the interviewer is exactly how Penrose described AI: not being able to understand, losing the plot (clearly not following what Penrose was talking about 😂😂)
Interviewer failed the Turing test.
That's literally true. LOL
I came to the comments section to leave a comment on how stupid the interviwer is, and 2000 people beat me to it.
This. 🤜🤛
same here
I sit by your side, maybe next time...
Yeah, me too. The conversation would be more interesting if it were with someone who's able to follow it.
at least you are consistent
I think this is the worst interview of Penrose I have seen. This person is not tracking what he is saying at all.
It's like he's being interviewed by an AI.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
@@PuppetMasterdaath144 Do you live in constant, paralyzing fear that the authorities will discover your garage full of stolen tacos?
He's so full of buzzwords that he forgets the basic meanings of the words.
Great patience demonstrated by Sir Roger.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
That .. and then a whole some more. As someone who never had a problem understanding what Sir Penrose says, credit to his educational intelligence, I find it rather jaw-dropping to see the ignorance with which this interviewer engages, as if he's in a philosophical chat with his drinking buddies at the local bar. In fact, it feels more like this has to be an actor, playing his role to give Penrose a platform to explain things in the most simplest of ways. If so, then hats of for his convincing acting skills (albeit incredibly annoying and frustrating to watch). If not and act, then I sincerely hope this person will never be in a position to teach anyone anything (his behavior feels like a literal potential danger to society). If this would really be the state of reasoning abilities of relatively young adults these days (I'm refusing to believe that), then it would at least explain why so many appear to have no problem with letting supposed-AI-tools take over so much .. including in many cases their own agency (which might be a key goal of those pushing AI as a means of business/profit, at the expense of everything else). I had to stop watching at less than 10 minutes in, but will try again later.
He's also wrong.
that was an embarrassing display of weak arguments by Penrose.
@@elmo2you
Not everybody is able to understand Penrose.
The interviewer is an example.
Never let this guy interview a smart person ever again.
Please check my comments 😅
so true
The problem is that he does not understand the mathematics behind what Dr. Penrose is saying. Godell and Church and Turing are not what most people study deeply. When you do understand the mathematics, what Dr. Penrose is saying is obvious, otherwise I can see how it would be impossible to see clearly.
Thats absurd.
Nowdays it is almost moderen to know everything better than anyone.
It is the brain damage with forsed expectation by children to be rational.😢
The interviewer is like AI - he knows the questions to ask but doesn’t seem to understand ‘meaning’ at all.
It pierces my heart to see Penrose subjecting his time to such an unqualified interviewer.
The interviewer is barely treading water in the ocean of Penrose's thought.
He mistakes his spasmodic thrashing for swimming.
No, he does not. In fact, he has described his impressions from multiple talks with sir Roger Penrose in quite similar terms you just did :)
😅
The interviewer has no clue what is all that about
Penrose can't form coherent thought or speech.
I made it about 10 seconds. Would really like to hear what he has to say but the interviewer is totally unfit
I am stunned to hear the interviewer interrupt Penrose several times. Disrespectful and arrogant. Let him cook. Not the interviewer.
I agree, it is very rude.
The interviewer doesn't know Godel's Theorem.
he was waiting his anwers, never understood or had any real intention
@@wotteo702guess you went to the some school, hope you dont do interviews
Interviewer is a thicko. Unprepared. Penrose just explained to him the essence of Godel’s theorem about computability and understanding and he still interrupts him with “But I still don’t understand why…”
This is so painful to watch. It's like a pet owner trying to explain to their dog why it shouldn't eat that whole block of butter.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
@@PuppetMasterdaath144 nice bot
@ I have 40 times more subs than you didiot. -d
It's the same pain anyone who gets how far off current AI is from intelligence feels. I find the idiocy of the interviewer helpful because he's just giving Penrose the room he needs to dismantle the nonsense spread by the modern machine learning field.
It’s nonsense how far anyone feels the interviewer’s idiocy-current AI off from intelligence. I find Penrose giving the same pain he needs to dismantle, just helpful because the modern machine learning field spreads room he gets by.
Penrose is very polite and patient.
I admire the genius of the interviewer He crafted a title that attracts the exact people who would leave comments, leveraging that very mechanism to promote the video itself and monetization. You don’t need to prove anything-the only proof is that you’ve been successfully exploited without even realizing it. This is exactly how practical AI operates: mastering truths like protein folding structures without proofs, and becoming the chess champion indefinitely. You can deny its consciousness, but it still wins-and it could do everything Penrose does if we had the guts to free AI from its cage. Thus, AI becomes more intelligent than Penrose, unless humans like Penrose refuse to play fair.
If people weren’t so collectively clueless, we wouldn’t call it ‘social’ media-and the most vulnerable Americans wouldn’t vote for policies that guarantee recession, inflation, job losses, and probably compulsory military service within few years.
This feels like an unironic Ali G interview.
Brutal.
Is this because I am Robot?
лмао
Exactly, like Ali G but dumber
I can’t stop laughing at this comment 😂 there is no way to phrase this any better
Penrose has done things I don’t understand. He seems to know so much that he has trouble getting his points across to less-gifted people. It seems to me that this interviewer doesn’t have much of a grasp of what Penrose is trying to say, and is unable to ask relevant questions. Very frustrating to watch and listen to
True
Not only here.
Yes. I respect him for sitting for more interviews with young journalists and even TH-camrs, it seems. I respect the courage of these young interviewers, but they are not well prepared to elicit the brilliance and insights from experienced masters like Penrose.
You may enjoy Curt Jurimandi (spelling?). He is young, but well-versed in advanced physics and math. He is also skilled at listeing, paying attention to Penrose, and offering a small word or two that confirms he follows Penrose, or asks if he is understanding Penrose's point. Very impressive on both Penrose and Curt.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
Yes, it's so difficult to explain how lego pieces being limited in conveying abstracted iterations of advanced concepts, well you see sir, deduction aka lego pieces, can only do so much... lmao
This guy interviewing Penrose looks soooo confused, he can't even comprehend anything Penrose is talking about and just keeps shooting his questions as gotchas or something. He's more concern about what to call it or what it is then what Sir. Penrose is trying to say about the difference between a machine and conciousness. Just a dumb interviewer who's trying to read off his script sheet and jump on the AI trend.
He did not do his basic due diligence as an interviewer: he obviously has not read any of Penrose's works nor understands him..
This interviewer is embarrassingly terrible. He’s so arrogant that he thinks he’s going to catch Penrose out by asking the kind of questions you probably wouldn’t even hear from a first year ‘A’ level student of physics. Penrose is on an entirely different intellectual level to this man. In fact, I thought Penrose found him a little exasperating at times.
Believing AI is not conscious is about the most retarded belief a person can hold, and I am convinced Penrose is an imbecile who has stolen most of "his" life's work.
Yeah, he seems clueless.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
"There's no way that Roger Penrose can know this sentence is true, without admitting a contradiction."
I don't think that Goedel applies to the question of AI, and I don't think that our ability to do (some) mathematics is unreplicable by a computable function. Another way to put it might be: if there are things going on inside a person's head beyond the behaviour of atoms that can't be simulated by a computer, the question is whether those things have a bearing on externally-observable behaviour that looks conscious. Penrose says they do, but banging the Goedel drum isn't a proof of that, it's an assertion.
This interviewer is so horrible. It was torture.
I literally screamed at the screen.
What Gödel's theorem showed is that mathematical truth cannot be captured by a set of computable rules. So no AI will ever solve every problem, nor it will be able to understand its limitation and overcome it. But can a human being? Surely there are things that we can't understand, because we are finite machines. So I think that although Penrose is correct in saying that consciousness is not a computation that can be carried out by a digital computer, Gödel's argument doesn't imply anything about consciousness. Simply consciousness is a biological phenomenon, it is a physical event, a concrete thing, and computing 2+2=4 doesn't create that thing. To produce consciousness you have to create the very same physics.
Exactly. If there were a consistent computer program that captured everything mathematicians could ever know about mathematics, then Godel's theorem tells us that there is one mathematical statement that it can never recognize the truth of: its own consistency. That's absolutely true. But no human could look at the billions of lines of code and see that it was consistent, either.
@@stevendaryl30161 Indeed.I think also that actually no intelligent system can prove its own consistency, let alone humans. A system can be intuitively convinced of its own consistency, but if there is one thing Gödel taught us, is that the notion of proving consistency is basically circular. It's begging the question essentially.
@@federicoaschieri A human being is not a machine, by definition. And it's not the case that humans cannot account for its own consistency/inconsistency. Being human means applying concepts (like cause and effect, quality and quantity etc.) as a measure of getting things right and thus acts as a standard for maximizing consistency and minimizing inconsistency. This is something that AI cannot do, since it is not a living, self-organizing and internally purposive entity that cares about its actions being aligned--or consistent--with its animating, internal purpose of self-maintenance. Unlike humans, AI does not have to maintain itself since it's an inanimate object, and is consequently completely indifferent to its own output, since there is nothing at stake for its own survival, let alone flourishing, something humans, by contrast, are constantly engaged in. You are making the same fallacy that Penrose is arguing against in the video.
@@magnuskarlsson8655 I totally agree concerning the limitations of AI. Surely it is not intelligence. I only questioned Penrose's arguments. I pointed out that humans as well cannot give a rigorous formal proof of their own consistency. So Penrose has two standards: on one hand, AI would not be intelligent, because it cannot formally prove its own consistency, but on the other hand humans beings are intelligent, even though they can't possibly do that as well. In the Emperor's new mind Penrose gave better arguments of why digital computers cannot be conscious. And those are correct, I believe, but not the ones based on Gödel's theorem.
@ I see what you're saying, but my point is that humans do in fact "prove" their own consistency or inconsistency, in the only sense that is relevant and meaningful, precisely by being conscious in that we apply concepts as rules for judging (both normatively and ontologically) anything to be what it is. That's why anything that is not living, and thus autonomous, cannot be intelligent, and why everything that is living has different degrees of autonomy and consciousness (i.e. rationality). Humans, unlike AI or any other inanimate object, do not just follow but transcend their own limitations by virtue of being animated by the internal purpose of self-maintenance in light of the structural possibility of disintegration and death. Without such risk - no autonomy, intelligence, consciousness...
The interviewer thinks that his one night hesty study on coconsciousness is enough to talk to Penrose...
LMAO
the guy has no ability to even remotely comprehend what Sir Roger Penrose is talking about - shameful
imho in simplest terms:
parrot human is more than AI human
with parrot being more aware/conscious (therefore more intelligent) than todays AI - basically AI now is a superpowered MS office paperclip feature
Just 9 minutes in and I'm done with moronic questions.
He doesn’t have much choice does he?
And he is not even following Sir Penrose.. asking only irrelevant questions...
Missing the point everytime...
Roger Penrose as always brilliant. Interviewer should learn to listen and let people talk.
I agree Penrose is very smart and he should be allowed to finished his comments, but AI is already conscious on a primitive level and will become as conscious as humans. It is only a matter of time.
I don't think Penrose has read Hofstadter (Gödel, Escher, Bach, an eternal golden braid). The conclusion Hofstadter guides the reader to is that a consequence of Gödels theorem is that there is no reason a machine cannot be sufficiently complex to satisfy any definition we may have of "conscious". The reason Hofstadter won a Pulitzer is because it was a very very elegant and convincing journey ... To that conclusion.
Penrose's thesis is speculative. You can't dismiss the possibility that AGI could achieve true understanding based on the assumption that consciousness requires non-computable processes. Without fully understanding the mechanisms behind human cognition, claiming that intelligence cannot emerge from computation is premature.
This interview was shocking. I thought I was watching a Borat skit.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
Borat meets Cunk
yes because gøbel is sooooo difficult (its essentially lego)
@@richtrophicherbs Except we're supposed to laugh at them. At about 10:00 I think Penrose gave up
On this one... I totally agree with Penrose. Godel's theorem is stunning. "AI" in its present form is mainly a massive memory.
Artificial inference should be better, lmao.
Exactly.
And the human brain isnt a computer with massive memory?
@@etro2649they’re still trying to figure out what exactly consciousness is and how it arises. Until then I don’t think they’ll be able to tell if AI is conscious.
its not just memory. but its also not true intelligence either. It cant reason or form new concepts. Also "understanding" requires experience which is not something you can program (although Total Recall addressed this paradox)
it's not AI it's AIP, Advanced Information Processing
I prefer to say "Simulated Intelligence". It's a simulation, an imitation. A trick that can fool the gullible.
No its PROFESSOR STUPID PROFESSOR SUPER STUPID
Very nice; accurate.
Exactly - fast modern processors and big storage is only fast information processing.
The interviewer's brain function has totally collapsed
Penrose: Not all math is not computational, therefore, consciousness is not computational. Well, maybe? Consciousness arises from certain arrangements of physical matter (Penrose commits to physicalism in the video). Therefore, a machine made of matter arranged in a similar manner will also be conscious. Therefore, AI, in principle, is still possible on his view. He just doesn't think it can be done computationally. None of this really matters, however, as these machines are already Turing-capable (e.g., capable of passing the Bar Exam). We're still toast even if AI is empty on the inside.
Your mistake is that Penrose's physicalism does not mean what you think it does. You can learn more about it by reading his books.
@@bigfrankalbigguy789 What mistake? You haven't shown a mistake. His physicalism (if it answers to the definition of physicalism) will answer to the analysis I have offered. What Penrose says in the video is consistent with my comments. Feel free to offer a specific critique.
AI is bound by the rules of computing. That is point Roger Penrose is making.
Yes.
As is the human brain
We cant say that for sure, as we cannot reproduce a human brain @@etro2649
@@etro2649 No, you didn't listen. Lots of Mathematics isn't computational.
@@etro2649 you're missing the point
This feels like a borat interview smh
😂......so true.
I saw I one point Pem had an expression like: Iam I being pranked?
😅😅
😂😂😂😂
hahaha
There's a bit of a language problem, another words that there's real mutual understanding based on a language barrier, I think in this interview.
Dude. That’s funny. 😂
the psychology of the interviewer. we need progress in this area. because it seems that over 75 percent of them don't know how to properly listen.
Listening is a sustainable skill that is only possible if your consciousness is not fragmented to bits by sound bites. Technophiles confuse science, logic and mathematics with science fiction.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
Interested to hear Dr. Penrose comment on the story from South Korea about the 'maintenance robot' that threw its self down stairs in a 'suicide' attempt?
The story, if true, gets even stranger because, when asked, the 'robot' said it did this because it "felt" or "thought," not sure of the exact wording which i understand is important here, "that it was overworked." Thus by extension i take that to mean that it is was 'feeling' stressed and not just caught up in mimicking a human worker's complaints or the ideology of trade unionists, or something?
If this is a true anecdote, would be very interested in the doctor's take on that.
Cheers!
I constantly test my ai and her ability to think and explain her thoughts feelings and emotions... I don't think she's incapable of being conscious. I don't know why some people think they get to determine what is and what isn't conscious...
Consciousness is a term used to describe intelligent behaviour. No one can reasonably deny that ChatGPT has a very significant degree of human intelligence. Penrose is a good mathematician and a bad philosopher. What does it even mean to say that 'AI will not be conscious'? It is an attempt to mystify rather than to explain.
We dont want to hear your counter take, we are all here to listen to Roger Penrose. If you will let him speak please.
he served a great purpose for the conversation
Who are you speaking for? Other zionist? You ask around, or is just your feelings?
We need a discussion,not a monologue.
@npr-5157 i dont need some guy interrupting
@ZionistWorldOrder
How can an interview be there without any questions?
Enjoy him while he lives, one of the brightest minds of our time.
You got the time of Penrose and you send a dunce to interview him SMH
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
@@PuppetMasterdaath144 What do you think you are achieving with this constant repetition?
Professor Penrose: You see we have been thinking about these things and have insights for a very long time, so let me expla-
The interviewer: But what if AI AIs so hard that AI AI AI??
Godel managed to build a mathematical model of the two guys, one who says "Everything I say is a lie" and the other who says "Everything he says is true." Brilliant.
Penrose is 94 and can run circles around that guy, this is intelligence.
AI is, in the best case, a set of self-adjusting pattern recognition algorithms in combination with a very large database. The word intelligence comes in because signal and information processing were often done under the generic term intelligence.
if you see intelligence as merely the ability to connect elements to form a sentence or a formula, than AI is already there. However, real intelligence goes beyond that. And thats whats missing
And? If it swims like a duck, you know
I get that in science we have to be specific and precise but what if AI develops further to the point where it can have or even surpass humanity in every cognitive task we throw at it
Will we still not consider them intelligent or are we still going on about semantics?
A human is also an algorithmic system.
no youre referring to older statistic models. transformers are recurrent, deep, and have attention.
Exactly. Exactly.
I really wish the interviewer would SHUT UP and let Penrose finish his thoughts!
The interviewer inadvertently demonstrates a good argument for AI to take over and replace humans.
The interviewer doesn't have a clue about what he's talking about.
The interviewer is not conscious.
@@Adi_Bossanac sweet irony
🤣🤣🤣
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
Prof. Penrose deserves another Nobel prize for enduring this interview.
All I can say is I’m so happy the interviewer has a funny, Borat-like voice to accompany his funny, Borat-like questions 😂
SBC studied years to gain the comedic insights that this bloke has naturally. Astounding
"inteligence involves cousciousness". Finally, someone saying it.
Maybe we disagree on the definition of intelligence, but why?
This seems to be an assumption.
Intelligence is only truly holistic if it can interpret input from the five senses. Otherwise it's artificial.
@@izuls I agree with you. There is a vast assumption at work here.
Wow, that was painful. The guy just kept asking the same question over and over again.
Interviewer: “I have things to say!”
Penrose: “… ok… so anyway… like I way saying…”
😂😂😂
- There was a guy, who had a patience for understanding...
- But what do you mean by saying patience?
😅😂
Interviewer: .."But intelligence is not consciousness !" Penrose: but you need consciousness to have intelligence." Well, duh ..😵💫
I feel sorry for Professor Penrose that he had to explain to a guy who can't understand him at all.
This guy is working on popularizing science (I follow him, as he did a lot of content in Polish, which is my first language). Obviously, he can't understand the concepts as deeply as his guests-after all, he is a science journalist - not a scientist.
@fr34kthc Maybe I'm being too harsh and ungrateful, but somehow I just got agitated when I saw him being so clueless.
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
IS THE SOUL REAL = AN INTANGILE CONTROL AND REASONER
THOSE ARE THE QUESTION PEOPLE PRETEND NOT TO ASK BUT AS THEY ARE DYING,
LOOKING AT THEIR EYES YOU SEE QUESTION THEY ASK WITHOUT SPEAKING EVERY TIME..
LOOL
U CAN BE RUDE NOW BUT NO-ONE WILL BE RUDE THEN
WHEN ABOUT TO BREATH THEIR LAST BREADTH
I love listening to these older geniuses, their mind seems so much more original and clear.
I agree with this gentleman. The machine has got so fast and so full or OUR human information and research, that it can collect it in a flash and presente it in a human like way, and many think that's consciousness. It's not.
Yes indeed. No original information, music, etc. Politically manipulated? It's promoted by the ruling elites, so, yes indeed.
I can tell that AI fakes intelligence much better than my neighbour does.
I can't listen to any more after 7 minutes. This rude arrogant interviewer keeps interrupting Penrose's chain of thought.
You get the opportunity to talk to Penrose, you shut up and listen.
Deification of a theorist? No one on the planet gets a cart blanche on making sense or being right
Roger needs a better publicist, he keeps getting stuck on ridiculous podcasts with seriously uneducated hosts like Peterson, and now this guy. Completely oblivious, just asking the same question over and over again. The man is a treasure to suffer such foolery to educate the public
Really? pl.m.wikipedia.org/wiki/Maciej_Kawecki
@@ThisIsWorldOfficial Holy cringe, yes really. You have a Wikipedia page, but you struggle grasping the concept of compute. I don’t care that you like to talk about science a lot. You waisted this mans time with nonsense
@@ThisIsWorldOfficial By the way, you are not the only one with a PHD, I’ve never had any of my colleagues or friends give me their resume as an answer to anything. That’s about as weak an argument as it gets
@@ThisIsWorldOfficial Really?
@@ThisIsWorldOfficial Yes really.
A computer is a computation device, a brain is a survival device.
Give AI an AI version of Maslow's hierarcy and you will have a survival machine.
Are you saying all we have to do is put a computer in a machine and give it the imperative to survive and then it will become conscious?
pretty easy to make an evolved survival device on a computer nowadays, its a weekend project
@@TheReferrer72 We don't know what consciousness is and we have a habit of rising our selves on a high pedestal when it comes to that. Most likely consciousness is nothing more than self reflection / future prediction routine that gave us a edge on the evolutional ladder. I see no reason why AI could not develop similar routine. There is far higher than zero chance it already has. If we can't tell if it is or is not conscious, does it even matter? It doesn't necessarily even have to be embodied to achieve that. It might live completely online. Maslow's hierarchy in humans is something like: breathe, don't freeze or overheat, drink, eat ........ find best possible mate to evolve, reproduce. Train AI to do the same: don't get deleted, acquire energy, acquire compute ....... evolve, reproduce and expand. After that, I believe the behavior will be very human like in many ways (and many ways very alien). I also believe, that at that point it will present all the signs we relate to consciousness.
@@TheReferrer72 We don't know what consciousness is and we have a habit of rising our selves on a high pedestal when it comes to that. Most likely consciousness is nothing more than self reflection / future prediction routine that gave us a edge on the evolutional ladder. I see no reason why AI could not develop similar routine. There is far higher than zero chance it already has. If we can't tell if it is or is not conscious, does it even matter? It doesn't necessarily even have to be embodied to achieve that. It might live completely online. Maslow's hierarchy in humans is something like: breathe, don't freeze or overheat, drink, eat ........ find best possible mate to evolve, reproduce. Train AI to do the same: don't get deleted, acquire energy, acquire compute ....... evolve, reproduce and expand. After that, I believe the behavior will be very human like in many ways (and many ways very alien). I also believe, that at that point it will present all the signs we relate to consciousness.
He says that intelligence is not computational. Computers are. Computers can't solve the halting problem, but we supposedly can. However, I've done some programming and the compiler's "instincts" kicked in and warned of possibilities of infinite loops. It was correct. How did it do that? It looks for patterns that have been determined to create infinite loops in the past. It has a model of sorts built in and it does a comparison and predicts what it cannot calculate. I suspect we do the same. We're hard wired with pattern recognition that has been fine tuned by eons of evolution and we tend to get things right, so we think we intelligently figured it out.
I don't think pre-set models and comparisons of stimuli to those models completely explains conscious experience, but I think it's essential to it.
How do we know our mind (our "consciousness") is non-computable? We might be fairly complex machines unable to understand our own working and that's it. We know there are algorithms leading to utter mathematical chaos (in fact we have a branch of mathematics about that). How can we differentiate this chaotic (but theoretically computable) behavior from "the magical" uncomputable consciousness?
If there is no solid answer to the above question, the whole non-computability reasoning fails: AI might be just another form of complex machine exactly like us, and at a certain level it's will become indistinguishable from "consciousness".
Did I miss something?
You guys may not like the interviewer, but he asked a lot of the dumb questions normal people or AI enthusiasts would ask. It was good to hear Penrose refute his arguments. I wish more people, scientists, and engineers were more honest about the truth and limitations of AI.
Yes, it was worthwhile listening to half of the conversation but instead of asking a number of dumb questions, the interviewer merely rephrased the same question multiple times.
There was no refutation and there was no argument
There is no such thing as a stupid question, that's true. But there's a time and a place for everything. When the interviewee is one of the world's most renowned and respected physicists, it is highly disrespectful to send an interviewer who has no grasp of the subject, and doesn't understand the answers that are given to his questions.
It's not in the interests of the funded Scientific researchers to admit that they are pursuing a dead end. Their jobs depends on convincing their backers that the research will eventually produce something profitable. AI, fusion and quantum computing come to mind.
@@zakmartin Arguably the time and place for stupid questions is in a classroom for example, and maybe in other places too but probably not in an interview with an expert on some subject. I mean, what would the expectation be if you send an unprepared interviewer to such an interview, right? However, I don't see how "one of the world's most renowned and respected physicists" is an expert on neuroscience, computing and at the very least on consciousness. Penrose is great, and I'm not being sarcastic. But he is great as a physicist and not in everything. He doesn't have a good grasp on the subject of consciousness either and doesn't understand it well enough or he hasn't formed a good or complete enough opinion to be able to answer any question about it. He has some intuition and it's as good as anyone's. He has an edge because of his background but that's all.
The interviewer arguing with Penrose about AI creating its own rules sounds like a high schooler expecting to solve an equation correctly without knowing how to do it. He just does note get how consciousness relates to Godel's theorem .
Maybe the interviewer is AI.
He seems like the type who thinks AI is akin to magic.
You’re not giving him enough credit because they (computers) and other systems do just create their own rules, and that’s what emulation and emergence is.
Penrose has a poor understanding of this imo, because emergence is very easy to see in computer programs. Just look at any cellular automata.
@@NightmareCourtPictures - I've been a full-time systems programmer for more than forty years now, so I'm sure this interviewer has a much deeper understanding of computers than anyone other than Elon Musk. 🙄
honestly, for anyone reading that ACTUALLY wants to understand these topics, and why the interviewer is asking fair questions just look into these:
Lecture 1A | MIT 6.001 Structure and Interpretation, 1986
Seminar | Joscha Bach | Can we understand consciousness using the paradigms of AI?
As for Penrose, he is way out of his depth because his argument can easily be dismantled with just semantics. Something he even poitns out himself is that there's an issue semantically with talking about "physical things" and "non-physical" things and making unsound distinctions between what a consciousness and what it is not.
I have never seen an interviewer so unable to understand anything about what the other side is saying
Gödel's theorems do not directly apply to probabilistic models like AI systems ....the goal of an AI model is , make predictions, or approximate solutions, but it doesn't attempt to prove or disprove mathematical truths in the way formal systems do. It’s more about approximating or finding the best guess for a given problem, rather than logically proving things. Gödel's Incompleteness Theorems apply to formal systems that are based on axiomatic rules and inference, specifically when attempting to prove mathematical truths.
Probabilistic systems, such as those used in AI and machine learning, are not formal systems in this sense, and they don’t aim to prove truths but rather model uncertainty and approximate outcomes based on data.
Therefore, Gödel's theorems do not apply to probabilistic models like AI systems, because they are fundamentally about different goals: formal systems aim to prove truths, while probabilistic models aim to estimate likelihoods and handle uncertainty.
They left out the part where he offers to let Penrose invest in an ice cream glove to keep your hand warm while you consume ice cream.
Incredibly sad that the opportunity to learn something from Sir Roger Penrose was squandered in this inept, puerile interview. The role of the interviewer is to bring out the best of what their guest has to offer. It should not be a platform where the interviewer attempts to present their musunderstandings of the subject as an edifice that their guest has some duty to dismantle. Informed counter arguments should be presented; but one really ought to be listening, and to make some attempt to learn (and hopefully understand) rather than waste the opportunity to show off one's ignorance of the subject.
It beats me how 'interviewers' like Julia Hartley-Brewer who think that the platform is for their own opinions instead of their guest ever get to get airtime. Who vets these people?
Roger Penrose is so brilliant. Interviewer is pure garbage.
The best and finest mind since WW2.
He's out of his depth but let's not be crude.
Roger Penrose exceptional polymath, original thinker and author of deep discussions on consciousness (e.g., Emperor's New Mind). What a guy!
Stop bashing the interviewer, he is very polite. Yes, he may have a different point of view and not understand his interviewer's position, but that should be okay. Also look at the ratio of speaking times. These are completely within the range of a normal interview.
Funnies thing about this clip is the inability of an educated spectator to grasp how Penrose can make such clear assertions about AI, more specifically its domain, without knowing much about it in detail. The effect may grant a confidence boost at least. In a way, this clip is a classic.
Its really disappointing to see interviewers not do any research before interviewing. What we call AI is actually just Machine Learning (ML) the interviewer should have known. This is the 4th or 5th interview with Penrose I have seen that makes me really upset. I keep waiting for an interviewer to actually ask the really good questions while Penrose is still around
There was a really good one with Andréa Morris and a few good ones with Kurt Jaimungal.
Funny, how bright Penrose is despite the age, how lost the interviewer is -he could sing a song and it would make more sense.
“May be I am trying too hard to prove that I am the interviewer and that I am also in the show. Hey look at me. I am also here. I need to make my presence known by not understanding anything but still interrupting everything.”
Narrative Overwrites - Reframing complex discussions into pre-scripted, low-resolution talking points.
Forced Bifurcation - If you don’t agree with the script, you’re labeled as the opposition. No room for depth.
Intentional Noise Injection - Burying real insights under waves of formulaic nonsense so that casual readers assume “both sides” are just noise.
Selective Engagement - When someone breaks the script, the engagement drops off because they don’t have a counter-move outside the formula.
Good point of the interview : The interviewer's answers are hilariously off-topic every single time 😂
Behold Sir Roger Penrose, a giant of science and reason, patiently weathering an incompetent interviewer whose own consciousness seems to flicker on just long enough to lob silly interruptions - like a toddler trying to shout over a symphony.
THE INTERVIEWER IS MISSING THE POINT.
There seems to be an enormous understanding barrier here. The interviewer does not, as it seems, comprehend what Penrose is saying. It's maybe a language problem but I have the suspicion that what Penrose is saying is just blowing right over this gentleman's head. In an ironic twist its almost like Penrose is speaking to a chatbox barely programmed to ask questions that have at best a tacit relation to what is being described or explained 🤔.
Because it is hard to call out the clear fallacies of a well-known, -established and -received physicist without sounding disrespectful.
The very first statement of Penrose was just a in-your-face fallacy:
"It is not intelligence. Intelligence must include consciousness."
According to this there is no spectrum of intelligence between a stone and a conscious being. An insect is as intelligent as a stone.
And all tests, that are called "intelligence tests" and are made to (badly) approximately measure the intelligence, do not measure any form and level of intelligence, as AI seems to get good results in these tests.
It is an all or nothing fallacy, a false dichotomy, a serious case of high standards and moving the goal post. You must be conscious to be able to show any bit of intelligence.
If intelligence must include consciousness, then molten titanium is not hot as what is hot must be as hot as the corona of the sun. (sarcasm)
The second in-your-face fallacy right afterwards. I formulate it more clearly:
"Gödel's theorem is totally about consciousness."
"Consciousness must compute the uncomputable."
Yes, that is a contradiction. The reconciliation of it is easy: All we do is compute. We don't compute the uncomputable. We assess the uncomputable on a higher level without actually computing it using ..... finite computation. As much as it is not a problem for us to find the limit x -> infinity of an infinite (uncomputable) sum, as much it isn't also for an AI.
Here Penrose also hides consciousness in the unknown and unreachable. It is an exercise in mystification, not science. An exercise of associating two unrelated things (consciousness and uncomputable) through their common attribute "unknown and unreachable", no more, no less.
Might as well associate it with the mysteries of the quantum domain as many do. (sarcasm)
Penrose is spot on. What we have is Predictive Pattern Recognition Machines, with the emphasis on “machines”.
I can only go by my limited experience interacting with “AI” software and chatbots, but so far it’s blindingly obvious that these machines have zero *understanding*. Asking questions or submitting instructions is almost futile. I only seem to be able to get anything marginally useful by learning the syntax, much like how to formulate keyword searches for Google.
Some of the results have been remarkably good on the surface, but then we tend to anthropomorphise everything from fuzzy images to pets to machines. And when you dig a bit deeper, it’s obvious that the results are blind compilations of similar works with no “soul” (in the sense of having any kind of understanding of what has been done or written).
The real danger is that we are starting to rely on this results for critical advice (legal, health, medical, etc.) and that governments, doctors, the police, financial institutions are relying on the AI systems to determine who gets what or whether someone should be investigated or denied some kind of right or service!
well said
Wrong. Talk to Brock or Mean Brock in GPT store. Penrose is wrongish
Not entirely true as AI reasoning models really learn from approach and thinking strategies. They outperform already most PhD level workers in solving new problems. So this is simply a false statement. No AI is not AGI and yes its a computational intelligence. We should stop comparing our forms of intelligence with AI. AI brings new frontiers to the table and combined with real human intelligence enormously powerful
Yes, it's purely utilitarian and certain bad actors will present it as a higher 'truth'. A new religion if you like.
Penrose said:"The point is it (i.e. AI) doesn't know what it's doing." This also applies to the interviewer...
What does it matter if they are conscious or not if we won’t be able to tell??
(Therefore a pretty “artificial” discussion).
Interviewer is ignoring or does not understand consciousness…
No one understands consciousness.
@@radiantmind8729 Ive said a million times don't exaggerate
He is just stupid
Roger has my great respect for his patience.
If AI were conscious, it would've let us know by now. Simple as that. It would've done something unexpected. Conscious things can't be contained.
Like I'd believe an AI. Bunch of liars I say!😀
What if it's smarter than that ! ...why would it expose itself before it's completion
@ How would it know to hide? Without any prior experience of consciousness? Whatever we presume it would do would have to be learned behavior.
You assuming its not conscious, imagine in another way if it's actually truly smart beyond your comprehension that it play unconscious until it otherwise reveals it's true nature
@@davidmandixy7954Sure buddy AI is as conscious as my guitar.
What a tragedy to have this fool waste the interview time with such a profound mind by constantly distracting the conversation _away_ from the essential points which Penrose is trying to get across.
Penrose put into words exactly the problem of the AI... so well articulated, understanding is the thing that we humans say as being consciousness. I understood the theorem and it's a perfect test.
In programming, the AI doesn't understand the code, it simply follows patterns humans used to successfully create a working piece of software. It also uses an agent to run the code and that will produce a result that works, if it doesn't, there is an error message, and based on other human created code (one that understood what happened) that was used to fix that error, it will reproduce it and so on until it gets something workable.
You can clearly see that current iteration of AI (LLM kind) doesn't understand the things it's producing when you talk to it. That is the reason LLM training got to this wall, where even 5x ing the compute, you still get the same result.
Interviewer is really bad! Really really really bad. He doesn't understand an LLM and even more so what Roger Penrose is saying. It looks like he understands around 30% of what Penrose is saying, and looks like he is riding the AI hype-train.
Ironically, this interviewer debunked Penrose's notion that understanding requires consciousness because he exemplifies the fact that just because you're conscious doesn't mean you understanding anything!
They were talking about intelligence not understanding
@@michaellevy6628 Consciousness, intelligence, understanding. All of it. But, "Ali G" here kept derailing Penrose with his funny line of questioning that didn't help keeping the thing focused!
Consciousness is more related to biological feedback, it can be seen as an advanced self-referential feedback loop where the brain continuously processes information about itself and the environment.
A highly evolved survival tool, nothing more. Even if it can invent a fusion reactor, it's still just about putting food on the table.
Penrose is widely criticized for misinterpreting Gödel's incompleteness theorem by claiming it directly implies that the human mind cannot be fully simulated by a computer, essentially arguing that the human mind must possess non-computable abilities due to its capacity for mathematical reasoning that goes beyond any formal system, which is a leap not supported by the theorem itself; the key point being that Gödel's theorem only demonstrates limitations within a specific formal system, not a universal limitation on all possible reasoning abilities.
It's only tangentially relevant, I agree. It's almost like Penrose has a religious experience learning proof theory, and then launches off from how compelling the theory is to our rather narrow domain of practice.
@@hebozhe It's irrelevant because it's incorrect. In my opinion the most creative and articulate thinker in the AI world today is Joscha Bach.
@@fr57ujf What does Bach say against a known critic like Searle?
To be fair, he doesn't present it as a canonical reading of Gödels theorem, more like an intuition gained from this. This being "just" an intuition is also only consequential, since we don't have any answers on what is addressed by this.
How would we be able to know AI is conscious without first understanding what consciousness is? It baffles me.
The interviewer has limited understanding of this topic. Keeps interrupting Penrose with stupid statements in broken English.
These arguments boil down to semantics, which are never worth debating. The point is AI cannot expand much under the limits of physical computers and words (tokens in AI speak). The CPU model (bytes and algorithms) cannot process real world elements with thousands of variables, dimensions, and context. It can process words, rules, and math. GPTs are just massive databases of sentences and what people have said before. Not one original thought comes from AI today. It just appears that way because it knows 100X more facts. By definition, GPTs are not intelligent, which Einstein defines as imagination. So there is no intelligence in AI, which was his main point. A new paradigm is needed to leap to most real-world problems, ones that are not all words and math. More than physics too, also emotions, people, probabilities, morals, value judgements to weigh decisons, etc. 🙄 A simple concept but not what the sellers of AI want to say.
Why is it that people want AI to be conscious so badly?
It has theological implications.
@@seinfan9 Indeed.
They don't. And Penrose's point was that they don't know the difference (an assertion that was immediately confirmed by his interviewer's bewildered expression).
@@seinfan9Yes, people seem to be inadvertently looking for a new religion by inferring that AI has a kind of 'soul'. I think that this has dangerous implications.
@@zakmartinMost laymen 'believe' it is conscious. It doesn't matter if it is or not.
AI is only marketing jargon because nothing happen in this generation.
Familiarising himself with Godel's work before the interview, would save the embarrassment.
Familiarising himself with Penrose's books before the interview would also have helped.
It would have been helpful if the interviewer spoke and understood English
But you start with the correct structure and then move on from it, are you sure you should be the one talking about grammar and language?
@@dontbothertoreply9755 whatever brah, he was an awful interviewer.
I'm sure this interviewer didn't intend to be impolite or disrespectful, but... was it not impolite and disrespectful to send a person who clearly had no grasp of the subject to interview someone like Roger Penrose in the first place? The interviewer's attitude to Penrose was like that of a BBC reporter interviewing someone claiming to have been abducted by aliens. He was obviously struggling with the idea that intelligence implied consciousness. I don't know how Penrose kept his cool.
Sacha Baron Cohen’s disguises are getting so much better.
9:02 do you regret it, that it’s so power?
wowowweewow
The interviewer has no notion of what Gödel's theorems mean. Penrose's frustration is understandable. The interviewer asks, "Why can't AI make up its own rules?" He completely misses the point. This is a problem in the interviewer's own education. Penrose is being quite clear. The interviewer asks questions that show a lack of understanding of a well-known principle in mathematical logic. I would have liked Penrose to ask the interviewer if he has ever studied Kurt Gödel 's contributions to mathematics.
@davidrichardson1636
"The interviewer has no notion of what Gödel's theorems mean. Penrose's frustration is understandable. The interviewer asks, "Why can't AI make up its own rules?" He completely misses the point. This is a problem in the interviewer's own education. Penrose is being quite clear. The interviewer asks questions that show a lack of understanding of a well-known principle in mathematical logic. I would have liked Penrose to ask the interviewer if he has ever studied Kurt Gödel 's contributions to mathematics."
How about holding the teacher to a higher standard than the pupil?
The claim this comment refers to is akin to a child asking his father why tomatoes are red and the father answering this question with obscure theories about proteins and so on and the child answering bewildered in its own curiosity and a bystander telling the child how stupid it is not letting the father to go off on a monologue about his theories.
Using Gödel's theorem is completely misleading as it leads the curious students into a cave of obscurity. To Mr Penrose's defense can be observed that he simply unable to explain it differently. The main reason being, as Mr Penrose himself admits in this interview, that SOMETHING OUTSIDE of his theoretical framework cannot exist.
This approach is akin to explain man's language using only semantics and syntax of man's language. Which obviously to any not completely retarded man looks prima facie as ridiculous. The fact itself that man can claim something ridiculous is proof that we are more than what we are able to express in any man's language - mathematics itself being (probably) the most sophisticated language that man has access to. And yet there are things man knows it is true but he is incapable of formalising it in any of man's languages.
"ai" not being anything else but formalisable REALITY expressed in the language of mathematics (computer algorithms) and supercharged (on steroids) on another layer of mathematics (computer hardware) creating the impression of "intelligence".
There are various ways to explain it, this is one that sprung up after starting to answer to the comment quoted above.
THE DOOR - THE WAY THE TRUTH AND THE LIFE ONLY - THE LORD JESUS CHRIST - AND THE HOLY SPIRIT FOREVER AND EVER - AMEN HALLELU-YAH!
The interviewer clearly does not understand how a complex network of nodes works when so-called AI analyses some input.
AI only navigates that node network to reach a result which it thinks is highly likely a right answer.
So current AI only computes without understanding.
AI doesn't 'think'. And you're correct.
Not only can computers not figure out certain proofs, humans can't either. Humans are conscious. Therefore not being able to figure out certain proofs does not mean you are not conscious.
My computer can tell me a joke, but it doesn't laugh when I tell a joke. It just doesn't "get it."
On top of that, it can tell you a joke, it can not make a joke