AGI in 5 Years? Ben Goertzel on Superintelligence

[@joshuasmiley2833]

Ben absolutely on the top of my list as far as best gurus in AI. I love his philosophy, relate to his passion, compassion, empathy and genius in the field. What a great representation of leadership towards the path of AGI.!

[@netscrooge]

Top of my list too.

[@andrewsheehy2441]

So funny. There will be no AGI - for so so many reasons. This person talks at such an abstracted level he would be better off as a sci-fi author.

[@joshuasmiley2833]

@@andrewsheehy2441 the truth is coming from a person like me that grew up without Internet. Everyone would be better off as a sci-fi author. We are living in a sci-fi world it just seems like Some people just refuse to open their eyes.

[@andrewsheehy2441]

​@@joshuasmiley2833 I've worked in tech for 30 years-both hardware and software-and I'm convinced that those claiming AGI (Artificial General Intelligence) is on the horizon cannot provide a scientifically robust definition of what it means to "understand." The entire field of AI lacks a solid scientific foundation because it isn't focused on explaining why a system behaves the way it does, or what underlying physics leads to the observed behavior. Instead, AI is primarily concerned with what the system does. This lack of emphasis on the "why" creates an unbridgeable gap between what AI systems can do and what we, as humans, can do. But it seems many in the AI community don't even want to consider that aspect.

[@netscrooge]

@andrewsheehy2441 Sorry, but your hardware and software experience is largely irrelevant. A background in complexity science would be much more relevant. AI is too different than the systems you have experience with. You don't call an electrician when the plumbing has a leak.

[@baz813]

"Humanity is not aligned with itself" profound!

[@ashleigh3021]

It’s gibberish

[@twinsoultarot473]

Humanity will be aligned with itself and its aims and its goals WHEN the focus is to take care of everyone on the planet. What more of a spiritual thing could there be? The biggest religions to unite people; has been their aim their purpose, their goal, but this would achieve the goal and carry out in a very realistic a well-thought out the most well thought out plan on the planet is for a resource-based economy and the reason why I say it's so well thought out is because everything is going to be tried and tested in research centers before it's rolled out on a worldwide basis. Yes, I'm talking about redesigning all of society, all of its mechanisms for distribution and for creation of goods and services. Think brand new smart cities ran by advanced AGI where everyone benefits instead of a few wealthy autocrats!

[@clarkflavor]

How would it look if we was?

[@Instant_Nerf]

@@twinsoultarot473horrible idea

[@TheRealUsername]

​@@clarkflavorThe end of war, global collaboration to move up in the Kardashev scale, faster economic growth and technological advancement, knowledge transfer among nations, AGI as a multinational long-term project instead of being carried by a few silicone valley gpu-rich labs with monopoly incentives

[@raybo780]

Mr Goertzel is hands down one of the smartest people I’ve ever come across. He’s such a warm guy too, what a great ambassador for AI

[@Paul_Marek]

Been following Ben since early 2000s. Glad to see he’s bringing some sense to the AGI hysteria.

[@chrism.1131]

Given his connection to China, I hope he helps the West achieve AGI before the East.

[@mrshankj5101]

i agree.

[@creatvsdd99]

I hope AGI comes as fast as possible to the public. My brother is struggling with mental health disorders which he isn’t getting help from any treatment. AGI can help solve issues hopefully:)

[@joecunningham6939]

He literally thinks ASI will solve all our problems in a utopian fantasy land. He's as hysterical as anyone.

[@Paul_Marek]

@@joecunningham6939 I don’t see how it couldn’t or wouldn’t eventually get us there? Maybe not full utopia, but far from the shitstorm we’ve created/are dealing with now. Star Trek is a good example of the potential.

[@ginogarcia8730]

Ben looks like the scientist that studied the alien spacecraft in Indepedence Day

[@swapticsounds]

Wasn´t that Brent Spiner?

[@ginogarcia8730]

@@swapticsounds yeah i realized it again after googling it lol

[@Iwantalloftheinformation]

I would say he looks equally unkempt.

[@keithmetcalf5548]

Gotta b from the pacific Nw. that hat, hair, n glasses.I'd Say Seattle or portland lol

[@johntanchongmin]

Fully agree, LLM is not the central component of the AGI system - it should only be one part of it, and a lot of the structure and data representation is equally or even more important than the LLM itself.

[@minimal3734]

I'm thinking OpenAI may be heading in the right direction. o1 originated from a language model. But after it had learned language it continued training with the sole focus on reasoning. So it seems that it is moving away from its origins as a language model and is developing into something new.

[@TheRealUsername]

​​@@minimal3734It's more like a robust algorithm enabler/emulator, the model has been trained to rely on specific template of reasonings that yield deterministically correct results using PhD-level tools, then it was continually bootstrapped on its reasonings with a GAN-like system (a discriminator and a reasoner), technically speaking, it's still an LLM but with a more natural Chain of thought implementation.

[@AIForHumansShow]

Love these interviews. Ben is such a fascinating person.

[@chrism.1131]

Agreed 100%. It blows my mind that evolution could take us to a point of such mastery. That we could be on the verge of creating something even more intelligent. We live in interesting times.

[@joshuasmiley2833]

@@AIForHumansShow absolutely such an amazing person and so personable I feel like he’s a friend amongst my friends

[@I-Dophler]

Ben Goertzel’s vibe is like a time machine back to the flower power days-love, peace, and enough hippy energy to power a VW van. It’s like he’s living in a world where tie-dye is still a major fashion statement and world problems can be solved with a guitar solo. You gotta love the optimism though, like he’s just one good meditation session away from bringing world peace. Makes you wonder if he’s got a stash of ‘70s vinyl hidden somewhere, ready to drop the next love and peace anthem!

[@THERE.IS.NO.DEATH.]

he's definitely got a stash of somethin' ☮️✌️🌼

[@I-Dophler]

@@THERE.IS.NO.DEATH. Weed....lol.

[@ginogarcia8730]

Machine Learning Street Talk is blowing my mind month after month

[@paxdriver]

Bump this comment

[@shawnfromportland]

Ben looks like Stoner teenagers' create-a-skater character on EA Skate3

[@Perry.Okeefe]

If youre trying to say hes great I agree.

[@shawnfromportland]

@@Perry.Okeefe the only AI researcher who knows what "grindcore" is 💯💯

[@mattwesney]

Straight to jail

[@arnaudjean1159]

This guy is a real human entity

[@ferologics]

eccentric as it gets

[@oncedidactic]

“Self organization in the activation space” - nice phrase. I love Ben’s way of saying “that’s true” about limitations, but also honestly granting fractional competency.

[@5GTrevor]

"honestly granting fractional competency." worthless word-soup. pat yourself on the back buddy.

[@oncedidactic]

@@5GTrevor I’ll make sure to bring this feedback to the next yt comment standards committee meeting

[@hubrisnxs2013]

​@oncedidactic ::boomshakalaka:: Thanks for stepping up for the community. If you wouldn't mind, I'd greatly appdeciate if you can point out that, given such negative language causes deleterious efficiencies in the culture, to be positive so that I'll be in a good headspace when later I download the internet

[@henrismith7472]

He looks like some kind of techno-shaman-warlock. He looks like he's staring into the future. I trust him to lead us through the singularity.

[@fpartidafpartida]

I was distracted for just a second by a conversation with my wife….I had to rewind (interesting word) for quite a ways to be able to catch up with his train of thought. Hugely intelligent and interesting scientist you are interviewing here. Thanks for making this available. Wish I could bring a cold 6-pack to this guy’s porch and just sit and chat with him about this stuff.

[@steve-r-lewis]

Dr. Goertzel brings so much clarity and wonderful insight to the AI conversation, thanks for getting him on MLST, awesome channel.

[@MixedRealityMusician]

Wow this was an amazing interview. I highly respect Dr. Goertzel's perspective, thank you MLST! As a leftie, the biggest concerns I've experienced at least in academia with transhumanism are 1. eugenics and the historical risks involved from an anthropological and sociological perspective and 2. The risk of dehumanization through technology. I love that the posthuman was brought up, however I really like the posthumanism philosophical framework as opposed to transhumanism in general as it doesn't have all the cultural and social baggage that longevity, anti-aging, gene editing etc. pose. It also directly addresses anthropocentrism. I also think aspects of transhumanism are almost religious. The belief that we will reach these awesome powers, immortality, life beyond death, curing all diseases, it all sounds a bit familiar within the Western cultural frameworks, doesn't it? But man do I deeply respect Ray Kurzweil's vision and his predictions. FM-2030 is also a hero of mine.

[@twinsoultarot473]

We just need to use advanced technology first for meeting people's basic needs and then we can explore all of these other things. The religions of the world also talk about caring for all living things.

[@kyneticist]

@@twinsoultarot473 *should. We _should_ at the very least do our utmost to ensure that people's basic needs are met.

[@mathewward6229]

Have we set the bar to determining AGI too high? The example of music genres not been able to be developed by AGI is asking too much, a single human couldn’t do it and it took many humans a 100+ years, countless failures, the development of new technologies and sounds before we can close to developing new musical styles and genres. We have moved the bar for accepting AGI to that of super-intelligence.

[@agenticmark]

Ben is one of the first people that meet all 3 metrics - kind - intelligent - honest most fail at least one of these and I cant take them seriously

[@crassflam8830]

The thing i think Ben and so many other get wrong is that they expect human-level AGI will be able to exponentially and rapidly upgrade itself, but this doesn't make sense to me (how do you know how to make yourself more intelligent without the labour of discovery; it implies you already have the capacity of that upgraded intelligence). As we know from LLMs, simply being faster doesn't grant you free knowledge from the void, you still need to do search (at apparently exponentially increasing costs for *linear* gains). Looking at it a different way, the fast-take off scenario implies that a given mind/brain/AGI (a system of given complexity) would be able to model/predict/comprehend a system of even greater complexity, but this just isn't tractable. Instead, I think what we get is a large organization of human-like/level AGI systems that cooperate together, just like humans do. Big organizations have big costs though, and they start to slow down. They will probably be incredibly powerful, just like how human organizations are able to do super-human things, but it will take time. Adding in the idea that learning needs to be grounded in some actual signal (the real world, ultimately), it seems kind of obvious that simply shitting out novel chip architectures, or model architectures, just cant happen unless you cover an exponentially increasing search space. I firmly believe there's no epistemic free lunch (ideas need to be tested), and testing takes time, real world interfaces/effectors, energy, etc... Things can only accelerate in the real world so much before the energy costs become too high (and rest assured, we need to move stuff around in the real world in order get that sweet sweet truth signal).

[@reinhardberger7083]

True

[@netscrooge]

False.

[@crassflam8830]

@@netscrooge False false.

[@netscrooge]

@@crassflam8830 "It implies you already have the capacity of that upgraded intelligence." You don't believe in evolution? Life can't come from non-life? Higher forms of intelligence can't come from lower ones? Emergence is a fairy tale?

[@TheIgnoramus]

You did point at one issue, energy. None of this matters if we can’t keep up with the energy demand for processing.

[@DavidJones-kz6ik]

Couldn't it be that LLMs are analogous to a "language center" of a human brain? In split brain patients, confabulations occur as the result of the language area of the brain lacking integration with other parts of the brain, due to the splitting of the left and right hemispheres. Similarly, LLMs confabulate (we call it a "hallucination" of the LLM) also because they are not constrained by being integrated with with, say, an analogous "frontal lobe" AI component. If we just think of LLMs as a language center, then we can understand why it seems reasonable that an LLM would just be one component in a larger AGI architecture - that's how we humans seem to do it.

[@HomesteadingAlaskatoMaine]

Dec 20, 2024 OpenAI announced 03 crossed the General Intelligence bench marks.

[@BrianMosleyUK]

Seriously, this interview has lifted my already sky high regard for the quality of MLST for intelligent AI discussion. Absolutely fascinating. Thank you again 🙏❤️

[@jald6277]

I didn’t know Zarathustra was Ben’s son. I love Zarathustra and learn so much from his reposts. Much love ❤

[@CodexPermutatio]

I like Ben Goertzel's general optimism. And that hat. Nice interview.

[@kriscrnomarkovic1109]

Ben does love that hat.

[@paxdriver]

@59:00 Ben woke me out of a dream with this: "I think the alignment problem is that humanity is not well aligned with itself"

[@SFJayAnt]

Hey, I love this guy but his voice sucks. No lie no cap all ❤️

[@RealShinpin]

People aren't libertarian. Thats the problem.

[@Iwantalloftheinformation]

Yeah, we are a divided planet of nation states with different cultural/historical backgrounds; and if we're feeding the A.I. all the data of earth, that would innately have to include elements of this division.

[@mohsincsv]

Loved the new PDF Show Notes!!

[@MachineLearningStreetTalk]

Thank you!

[@steve_jabz]

Love hearing from ben, always a lot of nuance and uncertainty that you don't get from the doomers, cynics or hype bros.

[@oncedidactic]

This

[@swapticsounds]

Ben, if you work with the "remix" function of Udio, and use that only slightly but iteratively with complex and changing prompts, it seems like it deviates more and more from the training data. Prompting gibberish in the manual mode is another way to increase the creative "heat" of the model. You´ll have to expect a lot of noisy cacaphonic results, but some stuff is always interesting. Maybe forward this to Ben if you are in contact.

[@stretch8390]

God, such a good discussion - the history of different fields, ethics, mathematics, comp sci, language, the list goes on. Ben's breadth and depth is quite astounding to me.

[@Drakmo79]

I see where Ben is coming from, but I would be cautious to compare an LLM in isolation to a Human and or Brain, because important parts are missing: embodiment (even if simulated), multi layer memory, inner agent interaction (like consciousness and inconsciousness), external agent interaction, external agent interaction with peers (other AIs), inner monologue. We first had L"L"Ms we then had Expert-LLMs, we now got true multimodal LMs. All of these missing parts feel pretty much doable in the next 10 years. Still I feel the LM would the driving force and heartbeat of this system. Let's look at his take on math. I am pretty sure if we give an MLM a large enough context to take all the maths class and the context weight is tuned right it would get easily the same level of abstraction as some of the best students. From there you can build up. One important step could be to "teach" base models and then finetune based on synthesized data of these in-context "teached" models for new base models. OR you use them for Reinforcement Learning from AI Expert Feedback. This evolutionary approach is more like the knowledge-based intelligence space developed in human society.

[@ginogarcia8730]

can't believe i get to watch this for free in this year - Ben let's us see into a great science future

[@Aedonius]

Not without extreme ads. I've never seen a channel with more ads. They are literally every 10 minutes

[@david_king_music]

To add to that tangent on audio invention and UDIO, it can be as profound as you need it to be, basically...but still with heavy human intervention... Regarding the leaps, you can use it like an experimentation engine to quickly generate different manifestations of the intersections of multiple genres, and you can be very specific/targeted. Presumably that Mozart + West-African drum fusion would be the first generation to get you to jazz - the "spark", so to speak. You could even prompt it in terms of creating a reasoning tree to get to jazz, mapping those creative leaps and cultural moments, or use an external second LLM, like GPT, to do so, and interact with it in that way. Of course it could have already created the next jazz but we wouldn't recognize it as such, until it was integrated in a cultural and music-subculture framework. You can also prompt it in ways that aren't genre based, asking it to symbolize a phenomen or element of the physical world or philosophical idea for example, And it will generate a groove, a morphology, and even create local events/"form" based on the idea. It definitely needs human intervention to weed out the anomalies, sequence them, highlight the interesting parts, perhaps isolate a melody or drum part in order to get inside the audio and edit it when necessary. But imagine that type of capability within a Digital Audio Workstation (DAW) where it could affect every aspect of the process...but as I mentioned earlier, that human intervention could be an external LLM or next-gen AI with a certain aesthetic or practical goal/motive... Now where it would get that goal is the question I suppose, as imo music and art in general arise from the interplay of the possibilities of toggling/choosing/manipulating/combining objective parameters with the expression cultural/societal/personal values.

[@DubStepKid801]

Great guest and great video as always. Keep up the great work.

[@shay5025]

Loved the references around 1:20:00 to Culture series books and also the Asian vs U.S. culture predilection toward positive or negative assumptions about AI and our relative tendencies for help or power seeking.

[@BrianMosleyUK]

I think this is the best interview I've seen with Ben, thank you so much for bringing so many ideas out. Wonderful interview.

[@DavidJones-kz6ik]

This is probably one of the more lucid (though always entertaining!) Ben Goertzel interviews; he really seems on point here. Thanks to MLST for always bringing out the best in their guests!

[@raybo780]

When is he anything other than lucid? Maybe over our heads but always clear as a bell

[@jeremiasengelmann7836]

Legendary statement to end it on. Humanity will either unleash Terminators, or reach eternal salvation, with little in-between

[@Iwantalloftheinformation]

I believe with happenings we are seeing right now. A.I. will cause us to mind fuck ourselves out of existence long before the emergence of physical robots with the intent to physically murder all humans.

[@chrism.1131]

33:30 maybe that is the key! Allowing it to create its own goal from the primordial soup would be a step towards consciousness.

[@chuston_ai]

I can imagine Geortzel and a bunch of super weird and amazing people sitting on the floor, smoking whatever, at 3:00am - listening to records - and discussing how sheafs in differential geometry might explain mental representation unions. That would be a good time.

[@therobotocracy]

I’m all over the idea of Genetic Algos and evolution combined with LLM’s!

[@arnaudjean1159]

Agree at some point of level intelligence it's an Inexorable pathway and a new paradigm

[@I-Dophler]

What even is going on here............lol? Is Ben quietly plotting a fashion coup, or did he roll out of bed this morning and decide, "Yep, today's the perfect day to rock the genius-skater-dude mashup look"? Either way, he's absolutely nailing that "I’m way too cool for your average wardrobe" vibe. One moment he's deep in thought, cracking the code to AGI, the next he's probably landing a perfect kickflip in his mind, already two steps ahead of the next tech revolution. You’ve got to appreciate the boldness of it all-it’s like his outfit is a walking metaphor for how he approaches life: unorthodox, daring, and always a few steps ahead. Ben’s look might just be as groundbreaking as the ideas he’s cooking up. Seriously, who else could make AI and skateboarding look like they belong together? Bold move, Ben, bold move!

[@michelstronguin6974]

Ben is right here, LLMs are kind of like word calculators. If you truly want to make an AGI you need to use the principles on which the neocortex does intelligence. Like sparse distributed representation, and motor neurons linked to receptor neuron across the whole system, which brings in the dimention of time. Jeff Hawkins did some good work on this in Numenta, but further work will uncover even more principles.

[@vivisector3489]

Such a great interview, appreciated.

[@shay5025]

Love hearing his thoughts - a great listen and a lovely recentering after all the recent hype.

[@pik910]

I base my trust in technical authorities in areas I lack expertise in on the size of their neckbeards or eccentric clothing. 10/10, solid heuristic so far.

For some reason, I'm getting more and more restless while waiting for the next episode🥰. Thank you very much, MLST crew, for the great value you're delivering

[@ragnarherron7742]

Follow the money. All organism compete by efficient self organizing. Human society organizes through value and competive markets. The unversal language of money is symbol for thermodynamic work.

[@TuringTestFiction]

Great interview. Always impressed with Ben and his thinking. Worth watching for the book recommendations alone!

[@I-Dophler]

The alignment problem in AGI development presents two interconnected challenges: humanity's struggle to unify around shared objectives and the inherent limitations of current AI technology. Humans, with their varied goals and values shaped by cultural, national, and organizational differences, find it difficult to establish a collective vision. This discord complicates the creation of AGI systems that can truly benefit society as a whole. The priorities programmed into these systems are often influenced by competing interests, which reflect the fragmented nature of human civilization. At the same time, while current AI models like large language models (LLMs) show promise in specific areas, they lack the capacity for the innovative, adaptive thinking that defines human intelligence. These models operate effectively within their training parameters but fail to push beyond into genuinely creative territory, such as pioneering new forms of artistic expression or tackling abstract problems in unprecedented ways. Overreliance on these systems could therefore limit AGI's potential to evolve into something truly groundbreaking. Ultimately, the challenge is not just about building an AGI that replicates human intelligence; it's about creating a system capable of transcending human limitations while addressing the wide-ranging, often conflicting needs of society. The future of AGI could lie in a paradigm shift-one where we design systems not merely to mimic human thought, but to reimagine and expand it, taking humanity into entirely uncharted intellectual and creative territory.

[@spencerwendt4800]

Totally on point...the bar is "creation" and "visioning"; something that does exist and bringing the creation to existence in a future time frame.

[@ThomasTomiczek]

AGI is superintelligence. I mean, even if development stops, hardware gets faster. Some years later your AGI is a swarm of AGI forming a discussion group and reacting in real time. And obviously - development will not stop.

[@dr.mikeybee]

Thank you. I'm learning a lot. BTW, we need to set the learning rate on individual nodes. When we get a subnetwork to an acceptable error rate, we need to shut off, or at least diminish the learning rate. I bet Max Tegmark could write an algorithm to benchmark changes to nodes. Of course one could start with an algorithm that uses distance from the central corridor of the weights where changes are statistically more frequent. This is where we can't get answers in RAG, from input documents in the center of the context window.

[@TBOBrightonandHove]

Excellent interview. Best view I have had into Ben's broader thinking on progress towards AGI, Transhumanism and Culture and surprising (to me) how much I accord with his ideas. His book now on my list. Thanks MLST!

[@Justashortcomment]

Well, let’s hear your new take on Jazz.

[@AAjax]

Yeah, that's my thing. Presumably all of us watching and discussing are GI. None of us individually will create an original new musical genre that other people care about either. When a new genre forms, it's in small evolutionary changes to existing forms, the sort he grants is possible with AI, combined with tons of human feedback.

[@JasonPruett]

In A.I. Artificial Intelligence (2001), Gigolo Joe’s quote “Combine Fact with Fairy Tale” highlights the blending of reality and fantasy, demonstrating how AI can merge different aspects to create something new.

[@petersoakell6950]

Thanks for sharing this. Ben always delivers)

[@I-Dophler]

Although Ben Goertzel’s discussion of AGI doesn’t directly mention Atlas Shrugged*, there are striking similarities between the themes of Ayn Rand’s novel and the efforts of AI researchers. Just as Rand’s characters in *Atlas Shrugged persevere despite societal pressures, AGI developers and innovators continue advancing groundbreaking technologies, often confronting skepticism and regulatory hurdles. Both highlight the concept of progress driven by individual creators who push boundaries and strive to revolutionize society, even when faced with opposition.

[@DJWESG1]

Only in that some of the people in this space are gated.

[@I-Dophler]

@@DJWESG1 I see the parallels too. The perseverance and creativity of innovators in AGI remind me of the characters in *Atlas Shrugged* who pushed forward against all odds. It’s that drive for progress, despite barriers, that fuels both.

[@DJWESG1]

@@I-Dophler yes, there will be sociopaths using this technology to further themselves in spite of the rest of mankind.

[@whyitmatters3076]

Man, your discussions are absolute bangers. I hope one day you can bring Shane Legg into the street. Thank you for such thoruoghly-researched and well-articulated discussions.

[@russthomo9602]

Interesting conversation. I haven't came across Ben before. Wish he let you get the whole question out each time first before jumping in most of the time though haha great thinkers seem to struggle to not jump in early on interviewer questions. Still thoroughly enjoyed the discussion

[@Anders01]

Interesting point by Goertzel that AGI will change the speed of the acceleration. Ray Kurzweil has excellent predictions, but maybe he extrapolated the history based on Moore's law etc when predicting a technological singularity by 2045. And even that is a very optimistic prediction according to many! Having SI happening sooner than that will be mindblowing. Could happen!

[@DJWESG1]

Yes, it will accelerate the crisis.

[@contactelisha8069]

'after you have super AGI benovolently coordinating things'....will it benevolently coordinate things? What is the evidence that benevolence is likely? And how will it be decided what benevolence might look like in that scenario?

[@oznerriznick2474]

Great discussion! What is the origin of intelligence? Is it.. 1- Extraterrestrials that are different than us but are trying to align themselves with us? 2- A creator who is like us but vastly more intelligent? 3- Our own intelligence as human beings? It’s a good bet that LLM’s or not, it may develop self-organizing, self-learning agents that could soon catch up with us. I’m not saying it’s a bad thing. Amazing, more like!

[@michaelstewart529]

You guys have been having on some great guests lately.

[@palfers1]

I love this guy :)

[@nisachannel7077]

Amazing, profound, super interesting

[@matthewclarke5008]

I don't think we will ever truly reach AGI in the literal sense because I was listening to a jazz pianist express the most exquisite music, he was in a trance and almost having an orgasm as he played, in the purest sense, and I could feel it radiate off him through the music, I really doubt AI could do that, that being said it will become very good at playing the things that don't go beyond the basic notes and dynamics, I just don't think that's enough for it to literally be able to do everything a human can do...

[@jay-p5n3t]

@Ben Goertzel do you have a copy of ray kurzweil's new book, any news or not worth reading?

[@Ikbeneengeit]

13:00 Jazz and other music genres didn't arise fully formed. They were the result of many different people iterating and adding individual small deviations to existing musical conventions. By building on eachother over hundreds of songs, new genres arose.

[@straizys]

Ray said its 16 years apart because AGI will require far less power than ASI , we will need to build hundreds of nuclear fusion power plants and that might take decades.

[@prannoyjitbanerjee4061]

Not if we use nanotechnology to power AI.

[@Paulo-ut1li]

Even being a transhumanist - or because of it -, Ben Goertzel is one of the few humans with a clear vision for a better future.

[@j05hau]

Love this guy

[@lavafree]

It shows he’s been at ai for quite a long time…all important concepts right and clear

[@jimj2683]

Large numbers of people working together is a type of super intelligence. The more people that are working on AI, the faster we will get to AGI/ASI. This is accelerating fast. There is a big difference between 10000 smart people working on AI and 1 million.

[@bossgd100]

thx for the PDF Show notes

[@robertzafaras4462]

and the greatest achievement in the AGI world is? The cost benefit of NVDA is? illusory . in reality what has AGI actually done? Please inform me. thanks

[@twirlyspitzer]

It's good to see someone not jerking us around with naivety exploiting promotional hype giving us the goods on the realistic timelines for for HL AGI.

[@Digital-Heresy]

Why the backpedal jump-cut at 54:54? Everything Ben was saying about libertarianism and hoping we can get something good out of the current situation before "The Great Enshitification" ensues and regulatory capture forces a dystopic headless slave worker future state was 1000% true! Just felt sad that y'all had to have what I imagine was an off camera "Hey we're all liberals on the left-coast, don't give those other guys too much credit and pay some lip service to regulation and big brother is ok, big brother loves us" talk.

[@123asilveira]

It is great to have thinkers like Ben. Since we know how our intelligence operates then we should take advantage of fueling AI of things that we are not good at it! And let AI do it for us! We should complement each other.

[@CriticalBaker]

What if huge short term memory leads to madness? What if distributed consciousness leads to hurd (=ant/bees) society? Do we want to become like that? Maybe each ant is "happy" with his life, but do we want that sort of "happiness"?

[@kowboy702]

To get jazz… you need a number of creative leaps… that come from one or more cultures… of people, not computation

[@darek795]

I wonder if we really could measure progress in AI research - if we don't know how it would work we don't know if we've achieved 70% progress towards AI or only 10%.

[@goranmajic4943]

Ben is 👍👍👍 its peak AI marketing and money making. And the mass is not understanding the and following. Great moment in history to watch and learn

[@jonathanduran3442]

We should train the models to know when to be precise and know when to be creative by adjusting their own temperature so something’s are less prone to “high probability” and more towards certainty. Idk, it makes sense in my head but I am no expert, just a common folk ramble 😅

[@Perry.Okeefe]

Feeling more for the suffering that's in front of your eyes is not incoherent, its adaptive. You are far less likely to survive and pass on your genes if you care equally for all suffering.

[@therainman7777]

That’s true, but it can still be incoherent in a more global sense. You seem to be saying if something is maladaptive from a survival standpoint, as a species evolves in a competitive environment, then it is globally, universally coherent and optimal. But that’s obviously not true. In ancestral times, it was highly adaptive to enjoy the taste of sugar, since it was hard to find and provided an immediate and concentrated source of energy. But here we are, thousands of years later, with hundreds of millions of people suffering and even dying from obesity and diabetes, largely caused by their totally “adaptive” and “coherent” desire for sugar. As we change our environment and circumstances, what is adaptive and coherent can also easily change.

[@Perry.Okeefe]

@therainman7777 Right, so it's not incoherent. It could be argued that it has become maladaptive, which I would argue is only true for the most privileged of human beings. Most of humanity still needs this adaptation. People with scarce resources need to be selfish.

[@oncedidactic]

There’s also an epistemic issue which is usually left aside- it’s much easier to assess your immediate local suffering with certainty. Consideration of suffering expands with perception of it, which is dependent on a lot of things.

[@gonzoz1]

Careful with you definitions. By suffering you mean human suffering , and the human gene pool is sufficiently small that care for distantly related humans makes sense for your genetic makeup.

[@squamish4244]

In modern terms, it's more practical to focus on the suffering in front of your eyes, because that you can get a handle on. If you can deal with that, you are more likely to be able to deal with more. Or all of it.

[@travcat756]

LLM's could be the start of a kernel that leads to AGI. Turn the monsters loose! Recursive self improvement will be the future

[@I-Dophler]

Imagine an alien race with five distinct sexes, each contributing something unique to the fabric of their society. This diversity would create a highly intricate and captivating civilization, where every sex plays a crucial role not just in reproduction, but in shaping the cultural, intellectual, and social structures of their world. The balance of power and cooperation would be essential for their survival, with each sex offering specialized skills or perspectives that enrich their collective experience. Their interactions would go far beyond mere biological needs. The synergy between these five sexes could lead to a complex web of communication styles-verbal, physical, emotional, and perhaps even telepathic. This would foster an unparalleled depth of connection, where understanding and empathy are key to maintaining harmony. Cultural development, too, would likely flourish under such diversity, with art, philosophy, and technology being influenced by the input of all five sexes. This could result in forms of creativity and governance that humans might find almost unimaginable, shaped by the interplay of their unique abilities and perspectives. In this society, problem-solving wouldn’t be limited to logic or reason alone; it might involve a rich combination of emotional intelligence, intuition, and collaborative thinking. The five sexes would likely come together to tackle challenges in ways that are deeply interconnected, where one sex might handle diplomacy while another excels in strategy or innovation. Their entire way of life would be built on mutual respect and reliance, offering a strikingly different vision of what it means to thrive in a diverse, multi-sex species. This alien race could redefine our understanding of diversity, showing that complex cooperation among various sexes is not only possible but essential for a truly advanced civilization.

[@ggrthemostgodless8713]

Is it an exponential curve or a power law...?? I think they are describing this in the wrong term, it seems to be a power law, not exponential, but it seems like that at the beginning, like NOW.

[@lutaayam]

Some people won’t believe AGI is here until LLMs are embodied

[@diga4696]

Intelligence brings people together. 45:21 let's get connected 🔗

[@marko-o2-h20]

Ray Kurzweil next?

[@lutaayam]

AGI was achieved last month

[@emma-lv7hn]

No

[@wuwv-m8h]

he looks very humble but is actually very very ambitious

[@suncat9]

Ben, 99.99999% of humans could not invent jazz, either, even if they're great musicians with a very high IQ, great musical training and experience. That's doesn't mean that particular human does not have general intelligence.

[@MarcoServetto]

What if LLMs are not the ramp and that is a good thing? That is, they do not have enough emergent reasoning to ever become dangerous in the world domination sense, but we can use them to help our goal quite a lot. In that sense, it would be amazing: just scale them up, make them more efficient and relax.

[@ragnarherron7742]

What is scary and disturbing is the pursuit of fantasy consciousness utopias. There is only the appreciation of each other and the privilege to pursue knowledge.

[@FORGOTTEN-THEORY]

9:42 what? x)

[@ADRIFTHIPHOP]

doesn't the war issue highlight how problematic it is for concentration of power where few make such decisions to put the many into those equations. the large bodies of leaders making collective descions for the greater good based on a multitude of perceptions is the way

[@Iwantalloftheinformation]

The Rain Man memorized I think the encyclopedia and dictionary, maybe dates to, and he could almost at just a glance count, or add up or perceive with perfect accuracy 100's of thousands, (Not sure about millions) of Objects in his visual perception field. The human brain is amazing. With some people's savant(I would even contend superhuman) abilities, its seems like the average human's brain wiring has huge hidden potential if it could at the structural level have its wiring reorganized a bit. As in most of us are born with this brain that, I guess for evolutionary reasons has this particular cookie cutout average range. But immense potential lies in just restructuring the brains we already have.

[@Iwantalloftheinformation]

There was also that guy that got hit in the head with a hard something, I don't remember what, and when he regained consciousness/health; He suddenly realized he was a math wizard, something he very much wasn't before. Or that condition where a person suddenly starts speaking in a foreign language/accent, even personality that have little to no experience with it.

[@MarkoKraguljac]

13:15 But music does not exist in a vacuum, its integral part of all other activities and physical life itself. Expecting AI to come from Mozart to rock music in a sterile vacuum, without wider context is just silly.

[@Adamhuehn]

The amount of power needed is over 750 nuclear power plants to power that AGI. Alberta Canada has a contract to build dams and nuclear plant sites and memory storage for it. Guess where I live. 😅 Not sure how to feel honestly.

[@pravdaseed64]

Thanks Ben. 💢 Magnificent "FM -2030" . 💢 "Transhumanist" He even change His Name for the Time @ that time.