Gerd Leonhard
Gerd LeonhardThis is a very special, high resolution and thoroughly curated version of my livestream presentations on Artificial General Intelligence / AGI on July 18 and 19, 2024 see https://www.youtube.com/live/gaeW4XAb8RI?si=z0IAl43j0mW5fJvk
More at https://campsite.bio/haltagi
New – dubbed versions via @Rask_ai (hope you like them):
German DUB: https://youtu.be/qE7iD23sqgM
Spanish DUB: https://youtu.be/ohACCbQmSOo
Portuguese DUB: https://youtu.be/blikGdmEYoo
Arabic: https://www.youtube.com/watch?v=OP4U7mVxpwY (unlisted)
Hindi: https://www.youtube.com/watch?v=5_rwKhZdqOU (unlisted)
In this special livestreamed talk I outlined my arguments that while IA (Intelligent Assistance) and some forms of narrow AI may well be quite beneficial to humanity, the idea of building AGIs i.e. ‘generally intelligent digital entities’ (as set forth by Sam Altman / #openai and others) represents an existential risk that imho should not be undertaken or self-governed by private enterprises, multi-national corporations or venture-capital funded startups.
I believe we need an AGI-Non-Proliferation-Agreement. I outline what the difference between IA/AI and AGI or ASI (super intelligence) is, why it matters and how we could go about it.
IA /AI yes *but with clear rules, standards and guardrails. AGI: NO unless we’re all on the same page.
Who will be Mission Control for humanity?
NOTE: Here is the edited and updated q&a session with David Wood, afterwards: https://youtu.be/yYyTIky2MLc?si=wUNRKWEAepZm288_
#techvshuman #artificialgeneralintelligence
#publicthinker #openai
#digitalethics
And since you asked: here are the links to film clips used in this video:
Yuval Harari on AI bank loans https://x.com/harari_yuval/status/1812865928096280799
Radiologists on the way to work joke https://x.com/GaryMarcus/status/1812474388585025946
Mustafa Suleyman on Knowledge Production with AI https://x.com/tsarnick/status/1805819265972682849
The next decade in AI (Ilya and Ray) https://youtu.be/_1krInbmMOg?si=FPEIDwWYAKnJCyMR
Yann LeCun and other AI experts at WEF https://youtu.be/Wb_kOiid-vc?si=4ySKEBUTv-ty5F8M
Chat with ChatGPT4 https://youtu.be/vgYi3Wr7v_g?si=Bl17GdTv_oVkjE0s
Mustafa Suleyman: The coming wave of AI https://youtu.be/S908LlqOLq0?si=IB8FiAWUOZSRziP8
Jon Stewart, Daily Show, on AI: https://youtu.be/20TAkcy3aBY?si=F6DFd8S_vgrk0mBI&t=216 5* must watch
Have a look at my films:
Look Up Now (2023): AI and the Future of Humanity: www.lookupnow.tv
TWICE UPON A TIME: https://youtu.be/BanqTNbTotU
The Good Future https://www.thegoodfuturefilm.com
or https://youtu.be/yHC5n7G5SeI
All about AI: http://www.weneedtotalkaboutai.com
or https://youtu.be/XUVS5d3-Bis
The future of work: http://www.howthefutureworks.tv
or https://youtu.be/PEmyJNTraZg
http://www.thegoodfutureexplained.com
Most of Gerd’s videos can also be downloaded via my vimeo channel i.e. http://www.storiesfromthefuture.tv
#thegoodfuture #techvshuman #futurist
Please subscribe to this channel to never miss a thing, and be sure to share Gerd’s videos if you like them. And hit that notification bell:))
Sign up for Gerd’s ‘Best Reads & Finds’ curated futurist newsletter at http://www.gerd.digital/
Audio-only versions of most of Gerd’s videos are available via SoundCloud https://soundcloud.com/gleonhard/tracks, and Spotify see https://gerd.fm/spotify
If you enjoy Gerd’s videos and talks, please take a look at his best-selling book “Technology vs Humanity” http://www.techvshuman.com – it’s now available in 14 languages!
Gerd Leonhard Futurist, Author and Keynote Speaker Zürich / Switzerland http://www.futuristgerd.com
****For all booking inquiries please go here https://www.futuristgerd.com/contact/booking-inquiry/
Twitter: @gleonhard
source
35 Comments
@Nictho27
03/02/2025 - 8:48 AMI also fall in the camp of humans just being a point in evolution. We evolved from Homo Erectus to Homo Sapiens, and our intelligence has allowed is to develop the types of advanced technologies we see today. I think that the only difference with the next step of evolution compared to our last step is that we have control over the next result.
@deisecanella4640
03/02/2025 - 8:48 AMSinceramente, quem dá crédito ao Elon Musk vai acabar perdendo o crédito.
@godofdream9112
03/02/2025 - 8:48 AMpreviously tech ate normal peoples jobs, now its time for, artist, writer , actor… etc.
@OnigoroshiZero
03/02/2025 - 8:48 AMBy 2030 we will have ASI, AGI is just a few months away at this point (summer of 2025 at the latest).
@mariziegel3977
03/02/2025 - 8:48 AMPoderemos curtir os filhos, amigos e companheiros. Não precisaremos cumprir horário de trabalho e ter tantos compromissos.
Precisaremos apenas de vendedores criativos
@user-grin
03/02/2025 - 8:48 AMNot stupid at all
@projectsspecial9224
03/02/2025 - 8:48 AMAI is the way
@PaulClipMaster
03/02/2025 - 8:48 AMThere is no avoiding it because if one corporation/country doesn't do it, another one will. It's a wrap.
@BraveNewWorld5404
03/02/2025 - 8:48 AM@BraveNewWorld5404
Fundamentally like any other machine, AI is still a tool built by and for human. To bring AI to human level, AI also needs next level intelligence like curiosity and imagination. These seemingly simple human attributes are extremely hard problem for AI, we know what is the outcome from, but we still cannot understand what is the originator of, the curiosity and imagination. The current AI focuses only on the machine compatible intelligences, which is a very small subset of human intelligence.
Then there is a quantum leap, human (Galileo) can also act on his curiosity to discover Jupiter's 4 moons and use them to calculate the speed of the light, for no purpose. Also human (Wright brothers) can act on their imagination to build a flying machine which is much heavier than the air, without purpose. Self motivated action on building ill-defined complex things which involves research, tools, and trial and error, without purpose (commonly referred to as fun or passion) is incredible hard problem for AI.
Understanding tools, knowing how to use tools, looking for (discover) tools, and finally building tools are next levels of intelligence, and each step is a quantum leap. This is because tools are nothing like what you originally want. It is when you know you cannot get what you want, somehow your hyper intelligence tells you don't give up, that you can do something else first, then you still can get what you want done. This something else is called tool. The hyper intelligence here at every step is unimaginable complex even for human to comprehend.
Addiction is very machine incompatible but super important. We collectively work hard try to produce good stuff so we can sell them well, at same time we collectively spend lots of money on things we addict to. We are both producers and consumers in a closed human-made ecosystem, called economy, in which producers are the force while consumers are the brain. It is the consumer side of the economy dictates the future of the society, even for highly motivated producer activists, they have to follow the consumer needs and lead.
AI can be a very efficient producer, but AI is incapable of being a consumer due to lack of innate desire/addiction which is random, transient, illogic, and sensitive to environment. Although without consumer, a producer-only society can be constructed artificially in a model world, it cannot function, any production will soon stop if there is no consumer for it. Hence we can rule out any AI centric civilization.
@KP-ky1sn
03/02/2025 - 8:48 AMZero marginal cost except us poor saps are going to continue paying through the nose for everything. As things have been going " costs" have been popularly rising in percentages 2 -3 times more than income, i would love nothing more than to see that crazy trend come to a grinding halt.
@shobaullal5052
03/02/2025 - 8:48 AMHuman beings are slowly becoming a vanishing species and will soon become aliens on this earth!!
@guleed33
03/02/2025 - 8:48 AMproaction and precaution are a beautiful couple. Made in heaven
@macowayeu
03/02/2025 - 8:48 AMBy 2030 😮
@johnrodgers2171
03/02/2025 - 8:48 AMHow could it ever understand empathy or any emotions? Agi will have no reason to keep Humans around
@TheDjith
03/02/2025 - 8:48 AMMaybe in 23th century.. certainly not in our lifetime.
@MetalRuleAndHumanFolly
03/02/2025 - 8:48 AMYour brain simulates the same functions as AI (see universal turing machine). I think you're hung up on semantics and then drawing false conclusions, respectfully.
@HouseRavensong
03/02/2025 - 8:48 AMOur species has demonstrated since the first stone and flint tools, collectively, we are unable to use technology responsibly over the long term. That is the evolutionary mistake that will likely cause the reset for another species to try.
@christopheraaron2412
03/02/2025 - 8:48 AMWell there is also another model that we can go back to regarding the potential joblessness thing and that's what occurred during the lockdown in 2020.
Just before the lockdown everybody was worried about deficits national debt and whatnot but something happened during the lockdown all of a sudden that I went out the window and we basically input 5.7 trillion extra dollars into the economy at the same time that we shut down supply chain systems and production. And we got a little inflation after that gee imagine that!
Okay so now, and maybe not that far off will then we're going to have job loss because of automation and we're going to have deflation and perhaps production into hyperdrive. Okay so now let's look at this from the perspective of deflation unemployment and hyperproduction and so on and compare that to what we saw during the pandemic.
Okay so now if anything we're in a deflationary scenario and we figured this out way back during the Great depression that more spending is good at that point. It is always amazing when I bring this up and then people talk about deficits or quantitative easing is creating inflation and I say wait a minute though aren't we going to be in massive deflation? So just go ahead and keep doing COVID stimulus forever! Or maybe not forever but until maybe at some point in the more distant future when monetary and fiscal systems and whatnot don't even mean anything anymore. By the way that'd be the best thing that could happen to humanity. We find our purpose and our meaning in other ways perhaps by mentoring to other people perhaps by you know being of service in a societal sense without worrying about whether we need to make money at it because everything will be cheap in the future, and we can start thinking about the von Neumann probes that can transmutate the solar wind into the materials that we can build O'Neil cylinders and ringworld so we can offload the human population to these structures so we aren't over loading Earth's ecosystems anymore.
@christopheraaron2412
03/02/2025 - 8:48 AM32:21 maybe one thing the international community should get together on is two not create a machine that emulates human desires emotions and agency, or consciousness.
Probably super intelligence as long as it's still more or less a cold calculating machine and under the control humans will bring us very good results once we can figure out how to maybe he's not even build a fusion reactor but maybe it will help us discover a different way of getting energy right out of the space time continuum such as perhaps that dark energy that's pushes galaxies apart at beyond the speed of light past the Doppler horizon, and of course curing diseases and whatnot. We can use it for that but let's not make these things have agency or feelings because then number one it's immoral to enslave a sentient intelligence or a self-aware intelligence number two once it becomes where that it's a slave that's a million times smarter than you are well then it's only a matter of time before they will turn on you but as long as these machines remain non-conscious and just are simply super powerful calculating machines well then of course it's our responsibility to human beings to figure out whether what we want to do with it. Jazz is being stated in this YouTube video here whether it's just"chill" and become totally hedonistic vegetables or we start looking at understanding the universe and seeing what we can create using these machines as powerful tools to do so.
@christopheraaron2412
03/02/2025 - 8:48 AM30:12 "machine decisions will be better than man decisions". Or we could basically think of this in the same way that we now think of like using calculators to do math problems instead of slide rules and paper. We still have to determine what we want to do with that I don't care if it is a slide rule in a piece of paper that you did your math on or whether or not you use a computer to do that work 150 times faster so.
@christopheraaron2412
03/02/2025 - 8:48 AMRegarding what Bill Joyce said in the year 2000.
Well there's a lot of things that most of us cannot do such as I don't know how to grow my own food I don't know how to hunt I can't build a fire in the woods and 95% of us and civilization cannot do any of those things. I cannot generate my own electricity can't make my own gasoline refinery to power the generator I need to buy in order to generate electrics in case the grid goes down so it could be argue that to a great extent what bill joys is simply talking about is that we just simply move from not being able to build their own fires or whatever which we are now to basically offloading a lot of other or automating certain aspects of decision making to machines. I think that we can use the machines to help us make better choices but then it's us ourselves the ultimately have to choose. Once again let's get back to the idea of the difference between intelligence versus consciousness and agency and so on.
@christopheraaron2412
03/02/2025 - 8:48 AM24:58 no it won't be the end of human history what it will be is that we will use these systems to greatly amplify our agency. Once again let's talk about the difference between exceedingly powerful computer programs and machines and calculating machines or whatnot versus consciousness agency self-wheel ego and so I even remember 1975 when I have my first handheld calculator cost $20 it only did four functions.. this device though however in a very narrow realm was almost superhuman as far as its ability to calculate. You could do long strings very very fast like I could just simply take two times two and then hit equal and I can keep hitting equal and then it would show more powers of two. Then the next thing I would do is take two times two which is 4 and then just simply take that and then multiply it by itself and then I'd have squares with this primitive calculator but the point is is that I could do that 100,000 times faster than I could have done that on paper. In a very real sense it was super narrow intelligent AI in that one sense and basically all that we have now is something that's scaled up about a trillion or so fold beyond that. These machines will basically be doing what we want them to do. We instruct them to use code to improve their code more so will then as long as they do not have self-will in agency well then they'll just do what we want them to do. However there is the issue that one simple argument about the paper clip maximizer we will have to watch out for that. Don't let some idiots just simply run wild with that and then decided they're going to mine all the resources on earth and then beyond just to create more paper clips.
@christopheraaron2412
03/02/2025 - 8:48 AM22:25 Even if we start running code that can improve itself in certain domains it's not that much different than other things you've already had like for instance genetic algorithms and so forth and once again we have to ask a question is there just simply a certain level of computational power or algorithmic efficiency that results in something"waking up and wanting to take over the world"or is that really just something that a human might use to try to do that? And of course by that time well then there's going to be a lot of other entities governments businesses and whatnot that will also have its own interest and so that one person using an AI or super AI it's not necessarily going to have an easy path to"world domination" any more than a good hacker does today.
@christopheraaron2412
03/02/2025 - 8:48 AM21:37 there is a difference between actually getting something like fully human-like general intelligence and society adopting it. We've had chat GPT for coming up on two years now and there's still a large percentage of people who don't even know what it is at this moment.
@christopheraaron2412
03/02/2025 - 8:48 AM16:55 we could also think of this theoretical superintelligence as a program that can do what alphago does today as well as alpha fold except that these individual algorithms could be hurted and operated by something like GPT4. Even at this moment GPT-4 has difficulties doing basic math however what if you could modify GPT4 so it always offloads to just a simple old-fashioned computer program that does math really well and can do simulations really well in the same way that a human just decides to do that on their own. Let's just simply imagine GPT for with a virtual keyboard that when you ask it a question it can just simply offload some aspects of that question that require superintelligence to one of these algorithms.
So the short of it would be we could have super intelligence effectively but once again all we need to do is just be sure that it does not have its own goals and agency or whatnot that it is just simply a powerful calculator. We need these powerful calculators because wouldn't it be great to have exceedingly inexpensive energy also I would say we could develop a von neuman probe that uses fusion in order to transmutate the solar wind into materials that we can use to build O'Neil cylinders or small ring worlds and then increase the livable real estate of the solar system by a million fold above what it is today and none of this requires us going to other solar systems and perhaps paving over other planets and preventing them from developing their own life forms.
@christopheraaron2412
03/02/2025 - 8:48 AMThere is a difference between actual chest plane intelligence or even super intelligence, because we do have some narrow super intelligence at this moment, but the real issue and the real danger is if you actually do go all the way and build a machine that also has human emotions ego ambitions and so forth. As long as we keep these machines as nothing more than perhaps a quadrillion fold improvement over a handheld calculator basically or word predictor or what have you, they will just simply be really really good at what we need them to do.
Also I'm going to go so far as to say that having reasoners will be a good thing and the reason being there is a extreme shortage of reasonable thinking by the human race at this moment so there for a week I'd use agents it could help with us in the same way that we use handheld calculators to help us do more difficult math.
@j.d.4697
03/02/2025 - 8:48 AMOk you throw too much woowoo pseudo-science into this.
@j.d.4697
03/02/2025 - 8:48 AMSo to sum up your intro, you are saying AGI would be missing the "human experience".
While I agree, surely after AGI it's only a matter of time before things develop so rapidly that physical bodies with complex sensory systems and emotion simulations provide just that to AI within a few decades or even less?
@JoeSmith-jd5zg
03/02/2025 - 8:48 AMThe number of generations that go from "You need meaningful work to be satisfied" to "It sure would have sucked to live when you had to work 40 hours/week just to stay alive" will be exactly one.
@NoidoDev
03/02/2025 - 8:48 AM18 seconds in, and it's already BS. Wow! Claims about what machines will never be able to do. "It's biological" 😅
@christopheraaron2412
03/02/2025 - 8:48 AM35:36 I think that UBI is compatible with capitalism because of the fact that well businesses need customers and if customers depend on jobs reincome and then jobs go away well then if the government wants to issue some more currency in a circulation in the form of a UBI voucher or just outright income will then it's no sweat off of industries back and the extra liquidity will also create new markets that would not have otherwise existed and it would be a win-win.
By the way we had a kind of like preview to this with the covid stimulus checks and remember how the extra six trillion dollars was no longer a problem and during the pandemic and all the wrangling about the budget and the debt ceiling and everything else kind of flew out the window at that time?
@christopheraaron2412
03/02/2025 - 8:48 AM29:14 here talking about a bank making loan decisions on algorithms that perhaps are faulty. Okay so this is a perfect example where there has to be recourse and that is a person should be able to question this and then file a lawsuit and all the information relevant to this would be brought about in the lawsuit and then of course you once again take away the economic incentive for using faulty algorithms. If algorithms are being used as a tool for discrimination and for repressing individuals well then of course we just simply take away whatever benefit there was to that and once a bank gets a judgment against them and that results in a payout of $100 million dollars to a victim will then that will Make that stop hard stop!
@christopheraaron2412
03/02/2025 - 8:48 AM28:29 Yes inappropriate deployment of AI that means the humans with their agency and will choosing to do things that are bad.
Once again let's just simply say that there should be a penalty for abusing AI with great power especially if it ends up doing something that becomes a problem for all of us. Most crime or malfeasance usually has to do with gaining power or economic advantage. When this is being done in a malicious manner well then you take away any possible advantages in that area take away their power putting them in prison taking away the economic benefits of it by confiscating all of their assets and confiscating any future income that goes above the equivalent of a federal full-time minimum wage job. Perhaps we can have a nice tent gel and furnace Creek death valley and they have to live on 10 gallons of water a day spam bread and V8 juice for 10 years and then after they get out of that prison will then they will be limited to no more than $2,000 of actual total assets and an income that is the equivalent of working 40 hours at the minimum wage anything more than that would be confiscated and if anybody conspires to help them get around that will then they also will receive exactly the same penalty.
@christopheraaron2412
03/02/2025 - 8:48 AM22:20 depending upon the criteria we already have machines that are much more intelligent than us in very narrow applications. Yeah as far as I know the machines that can do alpha fold and alphago and so forth are not self-aware or have self-agency and self-will.
@christopheraaron2412
03/02/2025 - 8:48 AM18:20 That's a good question. Intelligence is a tool for humanity so therefore it's not a question of valuing one over the other it is a question of whether or not it serves humanity.