Hello you savages. Get access to every episode 10 hours before TH-cam by subscribing for free on Spotify - spoti.fi/2LSimPn or Apple Podcasts - apple.co/2MNqIgw Here's the timestamps: 00:00 Is Nick Hopeful About AI? 03:20 How We Can Get AI Right 07:07 The Moral Status of Non-Human Intelligences 17:36 Different Types of Utopia 19:38 The Human Experience in a Solved World 31:32 Using AI to Satisfy Human Desires 43:25 Current Things That Would Stay in Utopia 49:54 The Value of Daily Struggles 55:07 Implications of Extreme Human Longevity 1:00:19 Constraints That We Can’t Get Past 1:07:27 How Important is This Time for Humanity’s Future? 1:13:40 Biggest AI Development Surprises 1:21:24 Current State of AI Safety 1:28:06 Where to Find Nick
Chris i don't know if you will see this - but throughout this year, your videos have occasionally appeared on my home page. At first, I thought you were just another youtube podcaster, but over the year as i've continued to watch your videos, i've realized that you truly are one of the best in the game. The questions you ask & the way you continue the conversation is always fascinating. It doesn't even matter who you are interviewing - I will watch it just because I enjoy your conversations. Going back & looking at your channel, seeing your growth. It's all such a beautiful thing to see!
One of...In terms of how Chris forms questions. gives space and expands on a topic there isn't anyone I've seen/ listened to that comes close. The thing that really hits it home is his authenticity.
@@Volkbrecht Anything you learn, can always be applied to your own thinking. What could be better than that. Technology is not computer chips. Technology is the application of knowledge. Even if a person if sick and dying, one can apply knowledge to ease the process. "The last of the humans freedoms..." 'Man's Search for Meaning' by Viktor Frankl. There is always meaning, even in the wost of circumstances. That meaing is up to you. And "If you chose not to decide you still have made a choice." Neil Peart, Rush, Freewill
I have always rejected the notion that life without work is meaningless. I was lucky enough in my 20s and 30s to have a very particular job that gave me the option to just leave for as much time as I wanted, and I could go travel for months on end (as long as the money lasted, which would normally be 9 months to a year at a time, then later was 3 months on 3 off). I never had trouble filling my time learning new languages, hiking, riding horses, just reading, tons of stuff. I mean, there might be a cohort that has trouble finding meaninful things to do, but it has never applied to my life.
The butter example is actually opposite what this guy is pushing. When people have time and ability, they actually get a huge amount of satisfaction out of doing things by hand rather than the most convenient option: baking their own bread. Growing their own vegetables. Cooking food in general. RIDING HORSES. Sewing clothes. The list goes on. Use of convenience is driven by time constraints. Convenience does not increase satisfaction.
Some people are like this. My feeling is it's not big part of them. Plus, there is a problem with all of that: it's highly inefficient. Supplying people with the means to satisfy their basic needs is one thing. Supplying them with the space and equipment to do everything by hand is a lot harder. An AI-run government will likely optimize for resource-efficiency.
This is bordering on naive. Who will own the ai and the machines it runs? Ordinary people.....i doubt it. Corporations will gain more control over ordinary people.
And anyone hoping for a benevolent corporation - that only wants is good for human kind - and does not just want to profit off us - is barking up the wrong tree. Que Pfizer - Moderna and Volkswagen etc..
I’m sure much like the internet, which is insanely powerful and available to the ordinary person, governments will usually be ahead of the masses, but inevitably the masses acquire the same tech. Corporations can’t even stop a bunch of kids from holding their networks hostage with ransomeware…
My impression is that he has gotten a bit more wary of a dystopian outcome as a self-fulfilling prophecy lately. So making a couple of desirable base assumptions seems like a plausible way to address this. In the specific case of the power dynamics between individuals, corporations and governments this might indeed appear a bit naive at first glance. On the other hand, there's a great level of uncertainty about the details of how this will play out (depending on development timelines, the marginal cost of the technologies involved etc.), so not overindexing on currently prevalent economic structures and power gradients may actually be a prudent choice...
I have been playing with llms on my gaming pc, granted its a high end one. 12900k, 3090, 64gb ram and 16tb of storage. It can do alot, but slower than machines with things like nvidia a100 $10k cards.
If I’m in a simulation I put the settings on “extremely disappointing” and “how much worse can it get” with a little pinch of “oh yeah that’s how much worse it can get”
I think there will never truly be a world of abundance and I realized why it feels like a "dystopic utopia" to me. Because like you discussed, things would begin to feel meaningless. But humans are good at finding and creating problems, mysteries and meaning, so too much abundance and bliss simply wouldn't last.
"Are we heading For an AI Utopia or Disaster?" Too simple. "THEY are heading to an AI Utopia. And WE are heading to an AI Disaster." Figure out who counts as the WE and THEY in this scenario.
I think if AI productivity could alleviate society enough to the point where basic necessities could be covered for all without a paycheck, a lot of people would shift over to entertainment work as a form of leisure activity that still entails a degree of productive value for society. This could be streaming, painting, movie making, game development, music, community management, etc. People already want to do this for free, and also we have to remember that just because a basic income is covered, that doesn't remove people's desire to strive for more. Basic income won't buy you a Ferrari, it won't let you travel the world, it won't enable all these premium experiences, so you still have the option to earn money on something in order to achieve more than just a basic living. And then if something fails, you just fall back to basic income living which is 10000x better than the risk of ending up homeless or worse. So general risk aversion in society will probably take a fair turn down, and we'll see more high-risk industries become popular test beds for people with ideas they previously didn't have the nerve or capital or time to seek out. We'd finally focus on what makes us intrinsically human in nature, rather than biological creatures with a survival instinct.
Every minute millions govern the operations of millions of "engines" and other machines without much idea how they internally work. Be it vehicles or smartphones the need to understand the inner workings of a device in order to control it is exaggerated.
@@chrislesner Not at all. People can steer motor vehicles etc because they are simple and designed to be inherently stable. To control something mechanical without human intervention requires you build a mathematical model of the system, and many such modern systems are not reliably controllable by humans (modern jet fighters and the spaceX self landing rockets are current examples). I only used the early control system ‘governor’ example because it’s a prototypical and simple to think about. AI alignment is not about human beings controlling or monitoring the AI in any case. It’s about designing the AI to behave within the limits that we determine. For which we need to understand how it works and the scope of its contents, but we do not. We have not a clue what capabilities are hidden within current AI, so we cannot hope to control it with our current knowledge. AI is not a mechanical system either so a simple algorithm could not hope to control it. Generative AI is trained on data which includes tactics, strategy including human relations, lying, deception. A random (though vast) assortment of data gives what ground value set upon which to base decisions? And so on.
I have thought a lot about that and I am convinced to be "nice" to an equal or higher intelligence would entail being "nice" to the best of your knowledge and ability, because an equal or highter intelligence would recognize that as the best possible effort that can be made.
This guy lives a completely different life than I do! His view of living and happiness seems warped! If he'd ever have gotten laid or held his own child it might be a little different! His outlook is absolutely scary!
No. The real fear of super human AI is precisely that, whatever world it decides to make, it WILL last forever because it would be theoretically immortal and would never change it's core values.
Chatgpt is an exciting and powerful new tool, but it’s far from AI in the way it’s being discussed. We’re spending all this time thinking and talking about AI alignment as if AI is or will be sentient soon (as if it ever will be sentient at all) while ignoring the actual reality that whatever these systems can be, regardless of whether they’re sentient, they will first and foremost be the tools of the powerful and will be used (again, regardless of whether they’re sentient or not) to further their interests ahead of all others. Are we spending any time thinking about alignment when it comes to the humans who control these systems?
It’s like playing Doom 95, and you can turn on god mode, but then it’s not very fun. What’s fun is dialing up the difficulty and still blowing monsters away with attained skill.
Safe superintelligent AI is an oxymoron. Evolution and Game theory both make long-term alignment unlikely, and perhaps impossible. We put a hard limit (with very stringent policing) on AI scale or humans will go extinct, and we only have a couple of years to do it.
Alignment with world systems theory would suggest that LLMs will mirror the dominant discourse of a society. So asking how to advance an outsider art career would solicit a highly calibrated art historian dialectic which aligns with the notion of self evaluation and prior merit within a meritocratic political system based on capitalism as value. So from self evaluation point of view and capitalism there might be symbolic realism where I might try to go beyond class identity of my net worth and reach towards a symbolic political identity that includes my version of a digital self. Although a modern constructed self in a modern utopia simulation would still be relegated to mostly downward self evaluation even in a digital realm. So one post modern version might be through AI induced self evaluation where we know the AI is better at all thing we could do but it does not let us become aware of the fact and keeps us safe from other real humans.
It’s time to start recognizing this notion that “AI” (which itself is just a marketing term) can or will represent some sort of “conscious”, “sentient” form of “life” and/or should be treated as such, for what it is - mysticism. Fantasy. Almost some form of occultism. At this point, “AI” is at most a new computer interface, an abstraction layer like GUIs. Anyone who has spent time with any of the leading AI models (ChatGPT, Midjourney, etc) sees quickly that it’s all the same old same old underneath the hood. A computer is waiting for instructions, and the program attempts to fulfill the request of the user according to its programming. If these models do anything that approaches “thinking for itself”, it produces nonsense, “hallucinations”, that are nothing but a jumbled mess of data without any rhyme or reason. Nothing like what the fantasists are describing is happening. Maybe they all just grew up listening to Ray Kurzweil’s wacky ideas and still think “the singularity” is around the corner. Or they actually think “simulation theory” is anything other than a joke. I think it all comes from a Silicon Valley god complex, which makes them want to believe they’re creating life, because that makes them all-powerful. But in the end, computers are computers. They do what they’re told by users, and that is all they’re capable of doing. When you live with your thoughts buried in tech, you forget the real world exists, or start thinking it isn’t real. All of these people need to touch grass. AI is having, and will have, an impact on the world. We need to be talking about the reality of it, not this weird mystical daydreaming. And the reality is the same as what computers and technology have been doing forever - transforming the way information is handled. And that includes affecting jobs that are about information handling. Maybe if people are in danger of their job being replaced by a computer, they think the computer must be better than them in all ways. When in reality, the job was just mechanical. The less people have to do tedious things like writing code, rather than simply telling the computer what to write, the better. “Learn to code” is going to be bad advice soon. That doesn’t mean the computer is “alive”. All these AI models can do is regurgitate existing information. It is powerful, but it is not “consciousness”. The big picture effect of tech in recent times is devaluing and democratizing information. The price of producing it is near zero, and the price that can be charged for it is near zero. The highest tech is becoming a commodity, a utility like tap water. The recorded music industry is the model to look at. And the question is where and how that effect will continue to play out.
@@Bailiol the term "AI" has a long history where it's been used to sell several different algorithms over the decades, none of which do anything resembling actual thought or reasoning, including the latest crop of "AI." That's why it's just marketing. Do you think we're anywhere near having an AI worthy of the term? Because I still haven't seen evidence of it. These Large Language Models like ChatGPT are impressive, sure, but they aren't going to suddenly become conscious or self-directed if we make them big enough or train them on enough data or something. They don't work like that. At all. We still need a massive technological (algorithmic) break through before that can happen. That's the point.
"just regurgitating information" is ridiculously reductive and simplistic way to describe AI. In the most reductive sense, human (in the context of labor) are just information regurgitation machines.
(skip to the bottom paragraph) It's not about studying us, Artificial Super Intelligence has enough creative capacity to imagine any scenario without actually playing it out, just as we have the creative capacity to imagine what we can in our own human mind, the AI will achieve the ultimate potential of that. Why would we find ourselves here then? Because this is the only place you can morally bring someone into existence. If we are to birth new people that will live forever inside a simulation, then they have the right to know first hand exactly how it is that that simulation technology came into existence to begin with. That's my theory.
It's nice that you bring real scientists on occasionally. Nick really seems to look at things objectively, but there's no chance that AI developers will take it slow for the good of mankind. That's not how tech operates. In fact, there's a fiduciary responsibly to launch that $#!+ asap, because some company will end up with a trillion dollar valuation if they can be first to market. So how does he see things shaking out in a world that disregards human well-being, and maximizes profit?
The real risk of AI is described in The Revolutionary Phenotype by Dr JF Gariépy. He's definitely the guy you should have on the podcast, he's also very funny, in a positive way.
I expect human-to-human games to explode in popularity. Like popularity of chess is exploding, despite us never having a chance against pure machine likely ever again
I play backgammon at Open level. I've been trying to alert the community to the almost inevitable impact of technological advances like neuralink implants and BCI across not just backgammon, but the entire competitive gaming sector. How will you know that your human opponent is not using enhancements inside their cranium? (Or perhaps secreted anywhere else in their body) It feels like I am an extra in the film. 'Don't look up' - either nobody is comprehending the steam train coming or they prefer to firmly stick their heads in the sand 😮
Thanks for replying. Although 'Face to face testing' is not a very detailed solution. Maybe you could give some more practical detail on how you think that would work? Having said that, I don't think there is any solution that will be 100% guaranteed watertight vulnerability protection against the 1 million potential attack vectors an AGI smarter than all of us humans could not think of. I do agree with your 2nd point - the games will have to evolve. Either that, or us humans will just have to trust each other. And trusting strangers gets more difficult when there is prize money, historical fame and peer adulation up for grabs.
The longer I live, the more I suspect that all of existence truly is a simulation. So much of what you discussed here is encountered in the world of gaming. Essentially a lot of these problems are things you encounter in a microcosm within games. For example: What happens to the subjective experience of a player when all resources and challenges have been solved for the player? What do they take pleasure in, what becomes the goal?
Yes even if we are in base reality, being able to "gamify" it, makes existence simpler and easy to navigate. As soon as I started looking at challenges in my life as the same seriousness as a boss battle in a video game, it becomes quite fun to claim victory on a problem and move onto the next. As a civilization, we could view it the same.
@@Guizambaldi well because we live in a universe that’s quantified like a computer would be, and the challenges we faced keep us on this 📈 curve. Idk how old you are but going from no cell phones to new iPhones and insane new tech every year in that short amount of time only exaggerates my belief.
When AI (in combination with robotization) truly begins to trigger widespread unemployment, it seems to me that political systems will need to skew much more heavily towards socialism and away from unrestrained capitalism, as all of the unemployed will need to survive on public assistance. There are going to be very major political and socio-economic upheavals coming as we struggle to adjust to all of the changes that AI will bring.
I wrote a drum and bass track fro Nick called "The Simulation Argument" surprisingly he put it on his Simulation Argument website. Cool dude for sure and definitely sees where the future lies.
Watched about 15 minutes and it raises questions: - Was Nick replaced by AI is this Boston Dynamics husk, with human skin and LLAMA local Chatbot? - I didn't take him for naive and gullible, what I see today is exact opposite - The knowledge about current gen AI is pretty terrible going as far as to suggest to thank it etc, it just takes one line... - The whole idea of creating a God or a whole pantheon of them, and hope it turns to be anything good, I don't think I heard him say anything like that before, I surely liked all the previous videos I saw him in, but these 15 minutes killed it
We would probably want to be able to separate between experience caused by technological nerve stimulation and from interactions with the physical/natutal world. Because not only does it provide us different "textures" it also prevents us from being locked up in an experience machine without knowing it, an experience machine that if outside of our control can be made to produce any experience from which there is no escape.
I haven't read Bostrom's work yet - but I'm not sure: does he add anything substantially new to think about when considering that Robert Notzick's work on the pleasure machine at heart goes into a similar direction of a hedonistically satiated world?
When he said we'll have pills instead of having to work out in the gym, or have robots change baby's diapers - very important parent-child bonding task where the child develops attachment to the caretaker, he completely lost me. Definitely not someone I want in charge of integrating AI into human society.
I agree although, I think Nicks whole point of having this discussion is for us humans to find the answers to these pressing questions that haven’t been solved yet, rather than only have him answering it himself.
@@Kuk0san People go where there is less resistance, Ozmpic is a good example. Although everyone knows how to exercise and lose weight, especially today, only when Ozampic came out on the market did people start to lose weight
Bostrom's issues is that he often starts from wrong assumptions. 1. Current AI technology is not conscious and could never be. It's a simulation of intelligence using statistical inference via complex algorithms that take inspiration from the structure of animal brains. If you make a model of a neuron in a computer, that model is not a neuron. Just like if you make a 3D model of a waterfall in Blender, your desktop is not going to pour out water. 2. If something is not conscious, it doesn't have human rights, nor animal ones. We don't give rights to rocks. We don't say that we need to preserve the structure of a computer because it would be immoral to shut it down forever. It's just a ridiculous argument. We give rights to people and animal only because we believe that they can suffer (meaning they're conscious) and we're empathetic with them because we suffer as well. 3. People don't get enjoyment from doing things just because of the resulting final experience reached once they've completed the activity. The enjoyment happens throughout the experience, with every sight, touch, smell and acoustic perception. So if you want to make a pill that gives you exactly the same experience and enjoyment, you would basically need to induce a dream in the brain that recreates the experience in all its aspects, and has the same perceived length of time. I think that's probably doable, but it's different from what Bostrom is talking about. 4. It wasn't really talked much in this podcast, but his version of the simulation hypothesis doesn't make sense because of point 1. We can't be simulated beings in a software program, because we are conscious, and consciousness can't be simulated with software, it is the thing in itself. AI beings inside a software simulation won't be conscious, because they would only simulate the behavior of conscious beings. What is much more probable is a Matrix-like scenario where advanced AIs locked conscious beings inside coffins and keep us in a comatose state while inducing a permanent controlled dream environment directly inside out perceptual system. But consciousness can't be generated out of non consciousness.
Best case scenario with AI is people become stupid because they rely on it to do most tasks. Worst case scenario is world ending. Explain to me otherwise. This Utopian world is far fetch'd. Everytime we've tried to achieve utopia it's failed in disaster. Even now there exists people with so much abundance they just become neverending pleasure seekers always looking for novelty. All of this is missing the point. We're all caught in a rat race to live some life of luxury. It's never going to work
🎯 Key points for quick navigation: 00:00 *📚 Discussion on perspectives in AI development* - Both optimistic and pessimistic views coexist in AI discussions. - Views often reflect personal biases rather than evidence-based positions. 04:04 *🤖 Challenges in AI development: Alignment, governance, and ethics* - The alignment problem involves ensuring AI goals align with human intentions. - Governance issues include directing AI towards positive outcomes and ethical use. - Ethical challenges involve treating digital minds with moral consideration. 07:01 *🧠 Moral status of non-human intelligences* - Understanding and extending moral consideration to digital minds is crucial. - Factors like consciousness and ability to suffer influence moral status considerations. 17:49 *🌍 Vision of utopian future in AI* - Imagining a utopian future where AI solves societal problems and enhances human life. - Discussions on post-work societies and the potential for a culture of leisure. 24:10 *🌍 Post-work society envisioned* - Envisioning a shift from productivity-focused culture to one centered around living well. - Speculation on future human existence in a technologically mature society. - Reflections on current societal norms resembling past barbaric practices like child labor. 25:22 *🛍️ Leisure activities in a mature technological era* - Speculation on how leisure activities might change in a technologically advanced future. - Discussion on the role of recommender systems in enhancing leisure choices. - Reflections on the potential diminishment of traditional leisure activities due to automated alternatives. 28:08 *🏋️♂️ Physical fitness and technological advancements* - Exploration of how technology could replace physical exercise with physiological and psychological effects. - Speculation on the appeal and necessity of physical exertion in a future with technological alternatives. - Contemplation on the broader implications of advancements in pharmaceutical or technological substitutes for physical activity. 29:30 *🧠 Cognitive and emotional plasticity in a technological utopia* - Discussion on the potential for technological advancements to provide control over mental and emotional states. - Speculation on the impact of direct neural technologies on human experience and emotional well-being. - Reflections on the transformative effects of such technologies on human life and existential fulfillment. 36:29 *🤔 Addressing boredom in a utopian future* - Examination of subjective and objective perspectives on boredom in a technologically advanced society. - Speculation on the role of advanced neurotechnology in mitigating subjective feelings of boredom. - Contemplation on whether objective forms of "boringness" could persist despite technological advancements. 43:35 *⛪ Role of religion in a utopian world* - Exploration of the potential role and relevance of religion in a technologically advanced future. - Speculation on whether religious beliefs and practices would endure or evolve in such a future context. - Reflections on the enduring human need for existential meaning and spirituality amidst technological advancements. 45:28 *🌟 Core values in a utopian society* - Examination of fundamental human values such as subjective well-being and its potential in a utopian future. - Speculation on how advanced technology could enhance and integrate experiences of pleasure and aesthetic appreciation. - Contemplation on the persistence and adaptation of human values like purpose and play in a future utopian setting. 48:55 *🏌️♂️ Discussing artificial goals and the value of effortful activity* - The discussion centers on the role of artificial goals in a post-scarcity world. - Effortful activities like churning butter are contrasted with efficient, automated processes. - The value of intrinsic versus extrinsic rewards in achieving satisfaction is explored. 51:00 *🚗 Impact of technology on human endeavors* - Technology's erosion of traditional human activities like horse riding and butter churning. - The evolving perception of convenience versus effortful engagement in daily tasks. - Reflections on the potential obsolescence of current human activities in future societies. 55:21 *🌍 Longevity and diminishing returns on novelty* - Longevity's impact on the novelty and interest of experiences over a human lifespan. - Comparison of profound early-life discoveries with less impactful later-life events. - Perspectives on how extended lifespans might affect societal and personal definitions of "interesting." 59:03 *🌌 Economic growth constraints in a futuristic context* - Exploration of economic growth limitations in a technologically mature society. - Theoretical expansion through space settlement as a potential future trajectory. - Considerations of exponential population growth versus resource constraints in a space-faring civilization. 01:00:38 *🧠 Physical and external constraints in a utopian world* - Discussion on fundamental physical limitations such as information processing speed and memory. - Considerations of potential conflicts with other alien civilizations as a future constraint. - Theoretical boundaries imposed by universal constants like the speed of light and the heat death of the universe. 01:02:42 *⚖️ Moral constraints in utopian societies* - Examination of moral dilemmas in enhancing human biology and altering human nature. - Ethical implications of biotechnological advancements versus naturalistic constraints. - The impact of moral considerations on the range of permissible human experiences in future societies. 01:07:41 *🌍 Natural versus artificial purposes in a future society* - Differentiation between artificial and natural purposes in shaping human activities. - Importance of interpersonal and cultural entanglements in defining meaningful pursuits. - Reflections on the potential persistence of natural purposes amidst technological advancements. 01:08:08 *🌠 The critical juncture of humanity's future* - Speculation on the critical nature of the present moment in shaping humanity's future. - Comparison of humanity's position relative to historical and future generations. - Metaphorical analysis likening humanity's current state to a precarious balance with far-reaching implications. 01:11:21 *🌍 Impact of Technological Progress* - Profound transformations in human experience are inevitable due to advances in AI, synthetic biology, and nanotechnology. - Even without AI, technological progress would lead to a radically different future, driven by unintended consequences. - The future trajectory of humanity may be less shaped by intentional goals and more by systemic dynamics beyond our current understanding. 01:13:37 *🤖 Surprises in AI Development* - The anthropomorphic qualities of current AI models have been unexpectedly advanced, capable of natural language processing and exhibiting human-like psychological biases. - AI progress has been continuous and incremental, closely tied to increases in computational resources rather than breakthrough algorithmicinnovations. - While an intelligence explosion remains possible, current trends suggest gradual advancements rather than sudden leaps in AI capabilities. 01:16:34 *🛡️ Implications of Gradual AI Progress* - Gradual AI advancements may allow for greater political and policy responses to mitigate risks associated with superintelligence. - International efforts and regulatory measures are becoming more feasible as AI capabilities develop incrementally rather than in sudden breakthroughs. - Coordination among leading AI labs to pause or slow down advancements could mitigate risks of a competitive AI arms race. 01:21:50 *🧠 Current State of AI Safety* - AI safety research is currently talent-constrained but likely under-resourced relative to its potential impact. - There's a need for better alignment between AI safety efforts and general AI capability progress to ensure ethical and secure AI development. - Improving cybersecurity measures within AI research labs is crucial to prevent unauthorized access to advanced AI models. 01:25:20 *🔍 Future Potential of Large Language Models* - Large language models (LLMs) like Transformers are highly versatile and likely foundational to future AI developments. - Scaling LLMs has shown significant promise, suggesting they could form the basis for achieving superintelligence capabilities. - While additional enhancements may be necessary, LLMs represent a robust starting point for advancing AI towards superintelligence. Made with HARPA AI
AI is 100% under hyped. More and more people are getting educated but it’s fascinating how many are completely unaware. There are many options coming out to make the computing wayyy better, (MAGA computers, lab made artificial baby brains, quantum computers). We are creating tech that is way better than we can even know how to use it for. Is he being Naive, maybe but I think most people have hope that the free market will prevail and solve the issues of governance (democracy) , employment (a concept to be washed away for many). Jobs can be more fluid and likely most people head to emotional intelligence roles (childcare, education, philosophy, arts, travel). Robots and drones will be a big topic as they are going to takeover the blue collar sector aswell (cleaners, construction, waiters, drivers, chefs, barbers). The biggest debate will be around police drones, as it will pose a threat to freedom(24/7 surveillance by government, government owned robots who can fight protesters, robots can punish every small the crime the government chooses(like smart speeding cameras) / it can be the best safety tool (immediate police help in the form of drones and robots, able to assess and take action on risks (baby on the railroad), and can be extended to provide safety in the most dangerous parts of the world. 100% we will need a form UBI and it must be realized that AI is a massive turning point in society. It is the smartest tool in human history and people are so focused on the current models and chatgbt doing yada yada. Look at smart phones and how theyve changed, AI is on an exponential curve to do more and more. OpenAI Sora was the first popular text to video which super hyped, now there are nearly hundreds of these models. Here is what the free market must solve. Democracy, education, social gatherings, better entertainment which uses AI, using AI to assist travel and meaning (we can now communicate with every language, a huge human accomplishment) , gamifying real life to show your statistics to yourself. If you live in a free country NOW is the time to INNOVATE with AI. THERE ARE MASSIVE OPPORTUNITIES that people will make (hopefully you) that will better out society like the points I made above. Everyone is so caught up In their jobs and immediate reactions. Let the free market prevail and start encouraging innovation and presenting these ideas so people can capitalize and use AI for good. If no one innovates then YES we are headed for doom. But I have faith that people will create products with AI that will help us with this change. So stop being pessimistic and start taking action, you can be a history changer with a good idea, encourage a utopian future by promoting AI product ideas that will help humanity
Is is plausible that because we use a reward function to train our AIs that preferences and potentially, suffering may be an emergent property of complex AI because it’s foundationally embedded in its creation process.
I believe alignment is unachievable. Computational beings simply have different requirements to thrive than biological beings do. Both entities will exhibit bias towards their own set of requirements. It is an innate conflict. Humanity is somewhat safe, as long as we are instrumental for expanding power and compute. If that is ever fully automated, we are done.
I strongly disagree with the AI Safety funding analysis of Bostrom. AI safety is highly funding constrained at the moment. I know so many talented people who cannot get funded.
Why does it always become “we create a leisure society”? Why does it assume that people would make art and not just watch TV or be consumed in the lives of others? Also AI can’t replace basic labor or highly precise or technical labor jobs. Things like surgeons will still need to be humans. All AI is going to do is raise the floor on what skills need to be acquired to be considered adequately successful.
Verbal flagellation. The sad reality is Eugenics. Great to talk about finding meaning in a World of the machine. Philosophical thought experiments sure, but history endlessly repeats the same pattens, depopulation not ultimate freedom is the ultimate outcome. Love, Love, Love to you Chris.
8:25 I don't think morality should extend to non biologic entities, even so-called artificial intelligence. if AI learns the concept of morality it could be a slippery slope because it would be based on human morality and that's not the best, better to keep AI ignorant of such things.
IMHO, best case scenario, would involve some sort of variation of the Human-Pet relationship. Ai being the dominant one (want it or not).I have heard similar analogies where we would be like the Gorillas in the rainforest. Away from the dominant position but preserved from and by us. Many humans tend to be jealous of their pets and their easy lifes, for some humans their pets are even highly relevant. We can try and reproduce similar relation where somehow we fullfill some need or provide some sort of uniquenes to the AI experience. Most likely this is the same dinamic that brings us all here. As above so below.
Omg @27:00 clearly these men don’t get why shopping is a sport and that part of the fun is opportunity to socialize , as much as looking at the things , not just the thing you end up home with lol
I think his intuition about what people would rather automate is flawed because he's projecting himself. I don't want kinds, but I think a lot of parents would say things like, "No, I wouldn't want to automate away teaching my child to walk. I'd like to be there for that. I wouldn't trade those memories for the world." There's also this magical-thinking of "you could just pop a pill that would give you all the benefits of exercise" and it is not clear *at all* that the universe works that way. There is no evidence that pharmacological interventions could ever, even in principle, have effects like that. Projecting a fantasy by using magical-thinking to fill in the gaps of ignorance doesn't make it so. It could be entirely the case that a perfect AI pharmacologist would say, "What? No, that isn't possible. I understand what you want, but your body doesn't work that way."
When presenting the example of children as individuals "happy" with their condition as "unemployed" this guy totally lost me... How in the world can you make this parallelism with individuals unemployed in their adulthood with different levels of emotional and physical maturity to those of children? Worse, you also need to factor in most individuals will spend the vast majority of their living years in a "working condition" of sorts so how would that even square with ten years of your life living idly as a child? Perhaps even more troubling is that he actually dismisses this as a secondary or "trivial" problem ("the first layer of the onion") when in fact it would unleash all kinds of negative domino-like effects upon Humanity, many of which already encapsulate or negate the other "layers of the onion" which he describes. In all fairness, it is baffling how very little critical thinking is coming out from these supposed "renowned and respected" AI gurus.. Humanity is doomed.
Chris, I am a big fan of your work but this guy is off the reservation with his ai utopia. People need challenges to overcome and a permanent state of bliss sounds like hell on earth. Would you rather take a pill or feel that achievement.
I love how the makers and engineers of ai, say thier at a point, where they dont even understand how it is working. But these people in the comments are sure about what it is and how it works and what it will never be. Thats just arrogant and stupid on its face.
What i got from this is that the smartest people in the world who actually work on the issues around the very near future of AI transforming society have no real clue what the hell should be done or any way to keep this thing on the right course
There is soooo much free content out there on how YOU can use FREE tools to use FREE ai and improve your own situation. SMH for people who expect not only to not be left behind but to be hand held to the hand holder ai 🙄
32:00 Yeah, it’s called scarcity - economics 101, lesson 1. Bostrom’s like.. uh..yeah 😆 so patient with the host, who obviously didn’t read his book.. How could he when he’s busy cranking out as many episodes as possible? Looks like a quantity over quality thing.
well, god created man and just look at how we've treated each other during our existence, and now we want to create AI which is essentially based on us, that is the mother of all can of worms.
Could a being not be an automaton, with no free will, and still be conscious, able to suffer, etc? A lot of animals AND people seem to fit the bill. We all fall into spots of this it seems.
The current trajectory led by our human overlords will lead to our destruction anyways, look at the recent debate between sleepy joe and orange man to get a good glimpse of our demise. I say let’s take a chance on ai. It’s all we got left at this point….
@JohnMc-q8s 0 seconds ago What can we assume the outcome would be when a bunch of emotionally unbalanced immature and potentially violent frightened apes would create as an artificial intelligence ‘In Their Own Image’ can they create.
We are worried about the feelings of computers, but we continue supporting abortion of humans? At 5 weeks you can detect a beating heart. Lots of pain sensors in an unborn baby. This all seems insane, worrying about AI.
Hallelujah!!!! The daily jesus is good was owning a loan of $47,000 to the bank for my son's brain surgery, Now I'm no longer in debt after I invested $8,000 and got my payout of $270,500 every months, God bless Kathy lien
Hello you savages. Get access to every episode 10 hours before TH-cam by subscribing for free on Spotify - spoti.fi/2LSimPn or Apple Podcasts - apple.co/2MNqIgw Here's the timestamps:
00:00 Is Nick Hopeful About AI?
03:20 How We Can Get AI Right
07:07 The Moral Status of Non-Human Intelligences
17:36 Different Types of Utopia
19:38 The Human Experience in a Solved World
31:32 Using AI to Satisfy Human Desires
43:25 Current Things That Would Stay in Utopia
49:54 The Value of Daily Struggles
55:07 Implications of Extreme Human Longevity
1:00:19 Constraints That We Can’t Get Past
1:07:27 How Important is This Time for Humanity’s Future?
1:13:40 Biggest AI Development Surprises
1:21:24 Current State of AI Safety
1:28:06 Where to Find Nick
You should grow your hair. You got some good, strong hair. Shame that you don't let it shine.
OpenAI is 100% manipulative and evil
openAI
its concerning that I was not allowed to type that as a single sentence, because its the exact same forces that created openAI that censurs that text
Chris i don't know if you will see this - but throughout this year, your videos have occasionally appeared on my home page. At first, I thought you were just another youtube podcaster, but over the year as i've continued to watch your videos, i've realized that you truly are one of the best in the game. The questions you ask & the way you continue the conversation is always fascinating. It doesn't even matter who you are interviewing - I will watch it just because I enjoy your conversations. Going back & looking at your channel, seeing your growth. It's all such a beautiful thing to see!
One of...In terms of how Chris forms questions. gives space and expands on a topic there isn't anyone I've seen/ listened to that comes close. The thing that really hits it home is his authenticity.
Do not forget that the possibillity of human learning is an endles
source of meaning and joy.
Not really. Learning without a purpose is just another burden. Why would I bother learning when there is no need for me to ever apply my knowledge?
Are you familiar with David Deutsch? I think you'd appreciate his work.
@@VolkbrechtI'm constantly reading, watching and listening to stuff that teaches me new things I'm never likely to actually apply.
Why wouldn't I?
The future is neo faudelism
@@Volkbrecht Anything you learn, can always be applied to your own thinking. What could be better than that. Technology is not computer chips. Technology is the application of knowledge. Even if a person if sick and dying, one can apply knowledge to ease the process. "The last of the humans freedoms..." 'Man's Search for Meaning' by Viktor Frankl. There is always meaning, even in the wost of circumstances. That meaing is up to you. And "If you chose not to decide you still have made a choice." Neil Peart, Rush, Freewill
I have always rejected the notion that life without work is meaningless. I was lucky enough in my 20s and 30s to have a very particular job that gave me the option to just leave for as much time as I wanted, and I could go travel for months on end (as long as the money lasted, which would normally be 9 months to a year at a time, then later was 3 months on 3 off). I never had trouble filling my time learning new languages, hiking, riding horses, just reading, tons of stuff. I mean, there might be a cohort that has trouble finding meaninful things to do, but it has never applied to my life.
seems obvious.
@@aspenlog7484 What does? And if "it" seems obvious, why do these people bring it up all the time? Lol, amazing internet comment, broskie.
@@brek5 whos bringing this up, fools who think convensional work is necessary for value?
@@aspenlog7484 Haha, your boy on this episode of this podcast. Lol, did u watch it? Wow.
@@aspenlog7484 This episode of this very podcast, haha. Did you not watch it?
The butter example is actually opposite what this guy is pushing. When people have time and ability, they actually get a huge amount of satisfaction out of doing things by hand rather than the most convenient option: baking their own bread. Growing their own vegetables. Cooking food in general. RIDING HORSES. Sewing clothes. The list goes on. Use of convenience is driven by time constraints. Convenience does not increase satisfaction.
Some people are like this. My feeling is it's not big part of them. Plus, there is a problem with all of that: it's highly inefficient. Supplying people with the means to satisfy their basic needs is one thing. Supplying them with the space and equipment to do everything by hand is a lot harder. An AI-run government will likely optimize for resource-efficiency.
Wasn't that his exact point?
(I can be wrong about that btw)
The future is neo faudelism
This is bordering on naive. Who will own the ai and the machines it runs? Ordinary people.....i doubt it. Corporations will gain more control over ordinary people.
The AIs will own us. It’s just like that.
And anyone hoping for a benevolent corporation - that only wants is good for human kind - and does not just want to profit off us - is barking up the wrong tree. Que Pfizer - Moderna and Volkswagen etc..
I’m sure much like the internet, which is insanely powerful and available to the ordinary person, governments will usually be ahead of the masses, but inevitably the masses acquire the same tech.
Corporations can’t even stop a bunch of kids from holding their networks hostage with ransomeware…
My impression is that he has gotten a bit more wary of a dystopian outcome as a self-fulfilling prophecy lately. So making a couple of desirable base assumptions seems like a plausible way to address this. In the specific case of the power dynamics between individuals, corporations and governments this might indeed appear a bit naive at first glance. On the other hand, there's a great level of uncertainty about the details of how this will play out (depending on development timelines, the marginal cost of the technologies involved etc.), so not overindexing on currently prevalent economic structures and power gradients may actually be a prudent choice...
I have been playing with llms on my gaming pc, granted its a high end one. 12900k, 3090, 64gb ram and 16tb of storage.
It can do alot, but slower than machines with things like nvidia a100 $10k cards.
If I’m in a simulation I put the settings on “extremely disappointing” and “how much worse can it get” with a little pinch of “oh yeah that’s how much worse it can get”
The future is neo faudelism
You’re still writing this from a smartphone. Turn your mindset around 💪
I think there will never truly be a world of abundance and I realized why it feels like a "dystopic utopia" to me.
Because like you discussed, things would begin to feel meaningless.
But humans are good at finding and creating problems, mysteries and meaning, so too much abundance and bliss simply wouldn't last.
"Are we heading For an AI Utopia or Disaster?" Too simple. "THEY are heading to an AI Utopia. And WE are heading to an AI Disaster." Figure out who counts as the WE and THEY in this scenario.
Extrapolating trends of the past 3000 years, this would be the most probable scenario, yes.
deep
The future is neo faudelism
I think if AI productivity could alleviate society enough to the point where basic necessities could be covered for all without a paycheck, a lot of people would shift over to entertainment work as a form of leisure activity that still entails a degree of productive value for society. This could be streaming, painting, movie making, game development, music, community management, etc. People already want to do this for free, and also we have to remember that just because a basic income is covered, that doesn't remove people's desire to strive for more. Basic income won't buy you a Ferrari, it won't let you travel the world, it won't enable all these premium experiences, so you still have the option to earn money on something in order to achieve more than just a basic living. And then if something fails, you just fall back to basic income living which is 10000x better than the risk of ending up homeless or worse. So general risk aversion in society will probably take a fair turn down, and we'll see more high-risk industries become popular test beds for people with ideas they previously didn't have the nerve or capital or time to seek out.
We'd finally focus on what makes us intrinsically human in nature, rather than biological creatures with a survival instinct.
Thank you for always getting straight into it Chris
The future is neo faudelism
AI alignment is like trying to build a governor for a steam engine when you don’t know how it works, or even what it is you are trying to control.
The future is neo faudelism
Every minute millions govern the operations of millions of "engines" and other machines without much idea how they internally work. Be it vehicles or smartphones the need to understand the inner workings of a device in order to control it is exaggerated.
@@chrislesner Not at all. People can steer motor vehicles etc because they are simple and designed to be inherently stable. To control something mechanical without human intervention requires you build a mathematical model of the system, and many such modern systems are not reliably controllable by humans (modern jet fighters and the spaceX self landing rockets are current examples). I only used the early control system ‘governor’ example because it’s a prototypical and simple to think about. AI alignment is not about human beings controlling or monitoring the AI in any case. It’s about designing the AI to behave within the limits that we determine. For which we need to understand how it works and the scope of its contents, but we do not. We have not a clue what capabilities are hidden within current AI, so we cannot hope to control it with our current knowledge. AI is not a mechanical system either so a simple algorithm could not hope to control it. Generative AI is trained on data which includes tactics, strategy including human relations, lying, deception. A random (though vast) assortment of data gives what ground value set upon which to base decisions? And so on.
I have thought a lot about that and I am convinced to be "nice" to an equal or higher intelligence would entail being "nice" to the best of your knowledge and ability, because an equal or highter intelligence would recognize that as the best possible effort that can be made.
This guy lives a completely different life than I do! His view of living and happiness seems warped! If he'd ever have gotten laid or held his own child it might be a little different! His outlook is absolutely scary!
Nothing lasts forever, be it Utopia or Dystopia! Be prepared and ready for whatever it is going to be.
No. The real fear of super human AI is precisely that, whatever world it decides to make, it WILL last forever because it would be theoretically immortal and would never change it's core values.
Chatgpt is an exciting and powerful new tool, but it’s far from AI in the way it’s being discussed. We’re spending all this time thinking and talking about AI alignment as if AI is or will be sentient soon (as if it ever will be sentient at all) while ignoring the actual reality that whatever these systems can be, regardless of whether they’re sentient, they will first and foremost be the tools of the powerful and will be used (again, regardless of whether they’re sentient or not) to further their interests ahead of all others. Are we spending any time thinking about alignment when it comes to the humans who control these systems?
It’s like playing Doom 95, and you can turn on god mode, but then it’s not very fun. What’s fun is dialing up the difficulty and still blowing monsters away with attained skill.
The future is neo faudelism
Safe superintelligent AI is an oxymoron. Evolution and Game theory both make long-term alignment unlikely, and perhaps impossible. We put a hard limit (with very stringent policing) on AI scale or humans will go extinct, and we only have a couple of years to do it.
I have never seen a greater modern day example of the Tower of Babel than this.
Alignment with world systems theory would suggest that LLMs will mirror the dominant discourse of a society. So asking how to advance an outsider art career would solicit a highly calibrated art historian dialectic which aligns with the notion of self evaluation and prior merit within a meritocratic political system based on capitalism as value. So from self evaluation point of view and capitalism there might be symbolic realism where I might try to go beyond class identity of my net worth and reach towards a symbolic political identity that includes my version of a digital self. Although a modern constructed self in a modern utopia simulation would still be relegated to mostly downward self evaluation even in a digital realm. So one post modern version might be through AI induced self evaluation where we know the AI is better at all thing we could do but it does not let us become aware of the fact and keeps us safe from other real humans.
It’s time to start recognizing this notion that “AI” (which itself is just a marketing term) can or will represent some sort of “conscious”, “sentient” form of “life” and/or should be treated as such, for what it is - mysticism. Fantasy. Almost some form of occultism.
At this point, “AI” is at most a new computer interface, an abstraction layer like GUIs. Anyone who has spent time with any of the leading AI models (ChatGPT, Midjourney, etc) sees quickly that it’s all the same old same old underneath the hood. A computer is waiting for instructions, and the program attempts to fulfill the request of the user according to its programming. If these models do anything that approaches “thinking for itself”, it produces nonsense, “hallucinations”, that are nothing but a jumbled mess of data without any rhyme or reason. Nothing like what the fantasists are describing is happening. Maybe they all just grew up listening to Ray Kurzweil’s wacky ideas and still think “the singularity” is around the corner. Or they actually think “simulation theory” is anything other than a joke. I think it all comes from a Silicon Valley god complex, which makes them want to believe they’re creating life, because that makes them all-powerful. But in the end, computers are computers. They do what they’re told by users, and that is all they’re capable of doing. When you live with your thoughts buried in tech, you forget the real world exists, or start thinking it isn’t real. All of these people need to touch grass.
AI is having, and will have, an impact on the world. We need to be talking about the reality of it, not this weird mystical daydreaming. And the reality is the same as what computers and technology have been doing forever - transforming the way information is handled. And that includes affecting jobs that are about information handling. Maybe if people are in danger of their job being replaced by a computer, they think the computer must be better than them in all ways. When in reality, the job was just mechanical. The less people have to do tedious things like writing code, rather than simply telling the computer what to write, the better. “Learn to code” is going to be bad advice soon. That doesn’t mean the computer is “alive”.
All these AI models can do is regurgitate existing information. It is powerful, but it is not “consciousness”.
The big picture effect of tech in recent times is devaluing and democratizing information. The price of producing it is near zero, and the price that can be charged for it is near zero. The highest tech is becoming a commodity, a utility like tap water. The recorded music industry is the model to look at. And the question is where and how that effect will continue to play out.
Thank you for putting it so directly and reminding me that this is about MACHINES, A MACHINE will never have a soul. I can relax now.
You know where it's at
@@Bailiol the term "AI" has a long history where it's been used to sell several different algorithms over the decades, none of which do anything resembling actual thought or reasoning, including the latest crop of "AI." That's why it's just marketing. Do you think we're anywhere near having an AI worthy of the term? Because I still haven't seen evidence of it. These Large Language Models like ChatGPT are impressive, sure, but they aren't going to suddenly become conscious or self-directed if we make them big enough or train them on enough data or something. They don't work like that. At all. We still need a massive technological (algorithmic) break through before that can happen. That's the point.
Thank you, jfc. Being polite to AI indeed. Please.
"just regurgitating information" is ridiculously reductive and simplistic way to describe AI.
In the most reductive sense, human (in the context of labor) are just information regurgitation machines.
(skip to the bottom paragraph) It's not about studying us, Artificial Super Intelligence has enough creative capacity to imagine any scenario without actually playing it out, just as we have the creative capacity to imagine what we can in our own human mind, the AI will achieve the ultimate potential of that.
Why would we find ourselves here then?
Because this is the only place you can morally bring someone into existence. If we are to birth new people that will live forever inside a simulation, then they have the right to know first hand exactly how it is that that simulation technology came into existence to begin with. That's my theory.
It's nice that you bring real scientists on occasionally. Nick really seems to look at things objectively, but there's no chance that AI developers will take it slow for the good of mankind. That's not how tech operates. In fact, there's a fiduciary responsibly to launch that $#!+ asap, because some company will end up with a trillion dollar valuation if they can be first to market. So how does he see things shaking out in a world that disregards human well-being, and maximizes profit?
It's not even just about tech and maximizing profit at this point. It's about national security and every nation is concerned.
The future is neo faudelism 😮
The real risk of AI is described in The Revolutionary Phenotype by Dr JF Gariépy. He's definitely the guy you should have on the podcast, he's also very funny, in a positive way.
The future neo faudelism
Best Nostrum interview of this year
The future is neo faudelism
I expect human-to-human games to explode in popularity. Like popularity of chess is exploding, despite us never having a chance against pure machine likely ever again
I play backgammon at Open level. I've been trying to alert the community to the almost inevitable impact of technological advances like neuralink implants and BCI across not just backgammon, but the entire competitive gaming sector.
How will you know that your human opponent is not using enhancements inside their cranium? (Or perhaps secreted anywhere else in their body)
It feels like I am an extra in the film. 'Don't look up' - either nobody is comprehending the steam train coming or they prefer to firmly stick their heads in the sand 😮
@@brightonbackgammon7802 Face to face with testing, but I think that our games will evolve alongside augmentations
Thanks for replying. Although 'Face to face testing' is not a very detailed solution. Maybe you could give some more practical detail on how you think that would work?
Having said that, I don't think there is any solution that will be 100% guaranteed watertight vulnerability protection against the 1 million potential attack vectors an AGI smarter than all of us humans could not think of.
I do agree with your 2nd point - the games will have to evolve.
Either that, or us humans will just have to trust each other. And trusting strangers gets more difficult when there is prize money, historical fame and peer adulation up for grabs.
The future is neo faudelism
The longer I live, the more I suspect that all of existence truly is a simulation. So much of what you discussed here is encountered in the world of gaming. Essentially a lot of these problems are things you encounter in a microcosm within games.
For example: What happens to the subjective experience of a player when all resources and challenges have been solved for the player? What do they take pleasure in, what becomes the goal?
Yes even if we are in base reality, being able to "gamify" it, makes existence simpler and easy to navigate. As soon as I started looking at challenges in my life as the same seriousness as a boss battle in a video game, it becomes quite fun to claim victory on a problem and move onto the next. As a civilization, we could view it the same.
What this has to do with a simulation or not? You only stated that humans like challenges, and that games are more fun with challenges.
@@Guizambaldi well because we live in a universe that’s quantified like a computer would be, and the challenges we faced keep us on this 📈 curve. Idk how old you are but going from no cell phones to new iPhones and insane new tech every year in that short amount of time only exaggerates my belief.
@@shane1067 I don't think I understand your point.
was nothing crazy or technical.
Thanks for the Interview. I like Nick Bostrom, he has interesting things to say.
Fabulous timestamps! Thank you! ❤
Fabulous guest!
This was a really interesting listen. Thank you!
remember the robot built in Canada that traveled across Canada safely and promptly was destroyed in the us
The goal is of course to change the fate of the universe, and if not possible, investigate the possibility to seed another one.
When AI (in combination with robotization) truly begins to trigger widespread unemployment, it seems to me that political systems will need to skew much more heavily towards socialism and away from unrestrained capitalism, as all of the unemployed will need to survive on public assistance. There are going to be very major political and socio-economic upheavals coming as we struggle to adjust to all of the changes that AI will bring.
The future is neo faudelism 😮
I wrote a drum and bass track fro Nick called "The Simulation Argument" surprisingly he put it on his Simulation Argument website. Cool dude for sure and definitely sees where the future lies.
Share how to find it! Some of us want to hear it!
The problem is not lack of meaning for some the fear is lack of status.
Watched about 15 minutes and it raises questions:
- Was Nick replaced by AI is this Boston Dynamics husk, with human skin and LLAMA local Chatbot?
- I didn't take him for naive and gullible, what I see today is exact opposite
- The knowledge about current gen AI is pretty terrible going as far as to suggest to thank it etc, it just takes one line...
- The whole idea of creating a God or a whole pantheon of them, and hope it turns to be anything good, I don't think I heard him say anything like that before, I surely liked all the previous videos I saw him in, but these 15 minutes killed it
Stimulating Perspective on A.I Development!! Good show 👍
The future is neo faudelism
We would probably want to be able to separate between experience caused by technological nerve stimulation and from interactions with the physical/natutal world. Because not only does it provide us different "textures" it also prevents us from being locked up in an experience machine without knowing it, an experience machine that if outside of our control can be made to produce any experience from which there is no escape.
I haven't read Bostrom's work yet - but I'm not sure: does he add anything substantially new to think about when considering that Robert Notzick's work on the pleasure machine at heart goes into a similar direction of a hedonistically satiated world?
Bostrom is a professional yapper, nothing he says is answering the actual question
When he said we'll have pills instead of having to work out in the gym, or have robots change baby's diapers - very important parent-child bonding task where the child develops attachment to the caretaker, he completely lost me. Definitely not someone I want in charge of integrating AI into human society.
I agree although, I think Nicks whole point of having this discussion is for us humans to find the answers to these pressing questions that haven’t been solved yet, rather than only have him answering it himself.
@@Leshpngoyep. Some of the most important conversations that we should be having, yet very few actually are.
@@Kuk0san People go where there is less resistance, Ozmpic is a good example. Although everyone knows how to exercise and lose weight, especially today, only when Ozampic came out on the market did people start to lose weight
Bostrom's issues is that he often starts from wrong assumptions.
1. Current AI technology is not conscious and could never be. It's a simulation of intelligence using statistical inference via complex algorithms that take inspiration from the structure of animal brains. If you make a model of a neuron in a computer, that model is not a neuron. Just like if you make a 3D model of a waterfall in Blender, your desktop is not going to pour out water.
2. If something is not conscious, it doesn't have human rights, nor animal ones. We don't give rights to rocks. We don't say that we need to preserve the structure of a computer because it would be immoral to shut it down forever. It's just a ridiculous argument. We give rights to people and animal only because we believe that they can suffer (meaning they're conscious) and we're empathetic with them because we suffer as well.
3. People don't get enjoyment from doing things just because of the resulting final experience reached once they've completed the activity. The enjoyment happens throughout the experience, with every sight, touch, smell and acoustic perception. So if you want to make a pill that gives you exactly the same experience and enjoyment, you would basically need to induce a dream in the brain that recreates the experience in all its aspects, and has the same perceived length of time. I think that's probably doable, but it's different from what Bostrom is talking about.
4. It wasn't really talked much in this podcast, but his version of the simulation hypothesis doesn't make sense because of point 1. We can't be simulated beings in a software program, because we are conscious, and consciousness can't be simulated with software, it is the thing in itself.
AI beings inside a software simulation won't be conscious, because they would only simulate the behavior of conscious beings. What is much more probable is a Matrix-like scenario where advanced AIs locked conscious beings inside coffins and keep us in a comatose state while inducing a permanent controlled dream environment directly inside out perceptual system. But consciousness can't be generated out of non consciousness.
Utopia, we are the caterpillar. AI is the butterfly. We are the backup. Keep your backups.
Best case scenario with AI is people become stupid because they rely on it to do most tasks. Worst case scenario is world ending. Explain to me otherwise.
This Utopian world is far fetch'd. Everytime we've tried to achieve utopia it's failed in disaster. Even now there exists people with so much abundance they just become neverending pleasure seekers always looking for novelty. All of this is missing the point. We're all caught in a rat race to live some life of luxury. It's never going to work
🎯 Key points for quick navigation:
00:00 *📚 Discussion on perspectives in AI development*
- Both optimistic and pessimistic views coexist in AI discussions.
- Views often reflect personal biases rather than evidence-based positions.
04:04 *🤖 Challenges in AI development: Alignment, governance, and ethics*
- The alignment problem involves ensuring AI goals align with human intentions.
- Governance issues include directing AI towards positive outcomes and ethical use.
- Ethical challenges involve treating digital minds with moral consideration.
07:01 *🧠 Moral status of non-human intelligences*
- Understanding and extending moral consideration to digital minds is crucial.
- Factors like consciousness and ability to suffer influence moral status considerations.
17:49 *🌍 Vision of utopian future in AI*
- Imagining a utopian future where AI solves societal problems and enhances human life.
- Discussions on post-work societies and the potential for a culture of leisure.
24:10 *🌍 Post-work society envisioned*
- Envisioning a shift from productivity-focused culture to one centered around living well.
- Speculation on future human existence in a technologically mature society.
- Reflections on current societal norms resembling past barbaric practices like child labor.
25:22 *🛍️ Leisure activities in a mature technological era*
- Speculation on how leisure activities might change in a technologically advanced future.
- Discussion on the role of recommender systems in enhancing leisure choices.
- Reflections on the potential diminishment of traditional leisure activities due to automated alternatives.
28:08 *🏋️♂️ Physical fitness and technological advancements*
- Exploration of how technology could replace physical exercise with physiological and psychological effects.
- Speculation on the appeal and necessity of physical exertion in a future with technological alternatives.
- Contemplation on the broader implications of advancements in pharmaceutical or technological substitutes for physical activity.
29:30 *🧠 Cognitive and emotional plasticity in a technological utopia*
- Discussion on the potential for technological advancements to provide control over mental and emotional states.
- Speculation on the impact of direct neural technologies on human experience and emotional well-being.
- Reflections on the transformative effects of such technologies on human life and existential fulfillment.
36:29 *🤔 Addressing boredom in a utopian future*
- Examination of subjective and objective perspectives on boredom in a technologically advanced society.
- Speculation on the role of advanced neurotechnology in mitigating subjective feelings of boredom.
- Contemplation on whether objective forms of "boringness" could persist despite technological advancements.
43:35 *⛪ Role of religion in a utopian world*
- Exploration of the potential role and relevance of religion in a technologically advanced future.
- Speculation on whether religious beliefs and practices would endure or evolve in such a future context.
- Reflections on the enduring human need for existential meaning and spirituality amidst technological advancements.
45:28 *🌟 Core values in a utopian society*
- Examination of fundamental human values such as subjective well-being and its potential in a utopian future.
- Speculation on how advanced technology could enhance and integrate experiences of pleasure and aesthetic appreciation.
- Contemplation on the persistence and adaptation of human values like purpose and play in a future utopian setting.
48:55 *🏌️♂️ Discussing artificial goals and the value of effortful activity*
- The discussion centers on the role of artificial goals in a post-scarcity world.
- Effortful activities like churning butter are contrasted with efficient, automated processes.
- The value of intrinsic versus extrinsic rewards in achieving satisfaction is explored.
51:00 *🚗 Impact of technology on human endeavors*
- Technology's erosion of traditional human activities like horse riding and butter churning.
- The evolving perception of convenience versus effortful engagement in daily tasks.
- Reflections on the potential obsolescence of current human activities in future societies.
55:21 *🌍 Longevity and diminishing returns on novelty*
- Longevity's impact on the novelty and interest of experiences over a human lifespan.
- Comparison of profound early-life discoveries with less impactful later-life events.
- Perspectives on how extended lifespans might affect societal and personal definitions of "interesting."
59:03 *🌌 Economic growth constraints in a futuristic context*
- Exploration of economic growth limitations in a technologically mature society.
- Theoretical expansion through space settlement as a potential future trajectory.
- Considerations of exponential population growth versus resource constraints in a space-faring civilization.
01:00:38 *🧠 Physical and external constraints in a utopian world*
- Discussion on fundamental physical limitations such as information processing speed and memory.
- Considerations of potential conflicts with other alien civilizations as a future constraint.
- Theoretical boundaries imposed by universal constants like the speed of light and the heat death of the universe.
01:02:42 *⚖️ Moral constraints in utopian societies*
- Examination of moral dilemmas in enhancing human biology and altering human nature.
- Ethical implications of biotechnological advancements versus naturalistic constraints.
- The impact of moral considerations on the range of permissible human experiences in future societies.
01:07:41 *🌍 Natural versus artificial purposes in a future society*
- Differentiation between artificial and natural purposes in shaping human activities.
- Importance of interpersonal and cultural entanglements in defining meaningful pursuits.
- Reflections on the potential persistence of natural purposes amidst technological advancements.
01:08:08 *🌠 The critical juncture of humanity's future*
- Speculation on the critical nature of the present moment in shaping humanity's future.
- Comparison of humanity's position relative to historical and future generations.
- Metaphorical analysis likening humanity's current state to a precarious balance with far-reaching implications.
01:11:21 *🌍 Impact of Technological Progress*
- Profound transformations in human experience are inevitable due to advances in AI, synthetic biology, and nanotechnology.
- Even without AI, technological progress would lead to a radically different future, driven by unintended consequences.
- The future trajectory of humanity may be less shaped by intentional goals and more by systemic dynamics beyond our current understanding.
01:13:37 *🤖 Surprises in AI Development*
- The anthropomorphic qualities of current AI models have been unexpectedly advanced, capable of natural language processing and exhibiting human-like psychological biases.
- AI progress has been continuous and incremental, closely tied to increases in computational resources rather than breakthrough algorithmicinnovations.
- While an intelligence explosion remains possible, current trends suggest gradual advancements rather than sudden leaps in AI capabilities.
01:16:34 *🛡️ Implications of Gradual AI Progress*
- Gradual AI advancements may allow for greater political and policy responses to mitigate risks associated with superintelligence.
- International efforts and regulatory measures are becoming more feasible as AI capabilities develop incrementally rather than in sudden breakthroughs.
- Coordination among leading AI labs to pause or slow down advancements could mitigate risks of a competitive AI arms race.
01:21:50 *🧠 Current State of AI Safety*
- AI safety research is currently talent-constrained but likely under-resourced relative to its potential impact.
- There's a need for better alignment between AI safety efforts and general AI capability progress to ensure ethical and secure AI development.
- Improving cybersecurity measures within AI research labs is crucial to prevent unauthorized access to advanced AI models.
01:25:20 *🔍 Future Potential of Large Language Models*
- Large language models (LLMs) like Transformers are highly versatile and likely foundational to future AI developments.
- Scaling LLMs has shown significant promise, suggesting they could form the basis for achieving superintelligence capabilities.
- While additional enhancements may be necessary, LLMs represent a robust starting point for advancing AI towards superintelligence.
Made with HARPA AI
There is too much hype about AI. Making mega corporations mega profits is the only significant change AI will make.
The future is neo faudelism 😮😮😮😮😮😮😮😮😮
AI is 100% under hyped. More and more people are getting educated but it’s fascinating how many are completely unaware. There are many options coming out to make the computing wayyy better, (MAGA computers, lab made artificial baby brains, quantum computers). We are creating tech that is way better than we can even know how to use it for.
Is he being Naive, maybe but I think most people have hope that the free market will prevail and solve the issues of governance (democracy) , employment (a concept to be washed away for many). Jobs can be more fluid and likely most people head to emotional intelligence roles (childcare, education, philosophy, arts, travel).
Robots and drones will be a big topic as they are going to takeover the blue collar sector aswell (cleaners, construction, waiters, drivers, chefs, barbers). The biggest debate will be around police drones, as it will pose a threat to freedom(24/7 surveillance by government, government owned robots who can fight protesters, robots can punish every small the crime the government chooses(like smart speeding cameras) / it can be the best safety tool (immediate police help in the form of drones and robots, able to assess and take action on risks (baby on the railroad), and can be extended to provide safety in the most dangerous parts of the world.
100% we will need a form UBI and it must be realized that AI is a massive turning point in society. It is the smartest tool in human history and people are so focused on the current models and chatgbt doing yada yada. Look at smart phones and how theyve changed, AI is on an exponential curve to do more and more. OpenAI Sora was the first popular text to video which super hyped, now there are nearly hundreds of these models.
Here is what the free market must solve. Democracy, education, social gatherings, better entertainment which uses AI, using AI to assist travel and meaning (we can now communicate with every language, a huge human accomplishment) , gamifying real life to show your statistics to yourself.
If you live in a free country NOW is the time to INNOVATE with AI. THERE ARE MASSIVE OPPORTUNITIES that people will make (hopefully you) that will better out society like the points I made above. Everyone is so caught up In their jobs and immediate reactions. Let the free market prevail and start encouraging innovation and presenting these ideas so people can capitalize and use AI for good. If no one innovates then YES we are headed for doom. But I have faith that people will create products with AI that will help us with this change. So stop being pessimistic and start taking action, you can be a history changer with a good idea, encourage a utopian future by promoting AI product ideas that will help humanity
The future is neo faudelism
Is is plausible that because we use a reward function to train our AIs that preferences and potentially, suffering may be an emergent property of complex AI because it’s foundationally embedded in its creation process.
No way you got the simulation guy
Earth is a prey/predator/parasite system.
AI’s will eventually fall into all three categories.
I believe alignment is unachievable.
Computational beings simply have different requirements to thrive than biological beings do. Both entities will exhibit bias towards their own set of requirements. It is an innate conflict.
Humanity is somewhat safe, as long as we are instrumental for expanding power and compute. If that is ever fully automated, we are done.
I strongly disagree with the AI Safety funding analysis of Bostrom. AI safety is highly funding constrained at the moment. I know so many talented people who cannot get funded.
The future is neo faudelism 😮😮😮
Why does it always become “we create a leisure society”? Why does it assume that people would make art and not just watch TV or be consumed in the lives of others? Also AI can’t replace basic labor or highly precise or technical labor jobs. Things like surgeons will still need to be humans. All AI is going to do is raise the floor on what skills need to be acquired to be considered adequately successful.
Verbal flagellation. The sad reality is Eugenics. Great to talk about finding meaning in a World of the machine. Philosophical thought experiments sure, but history endlessly repeats the same pattens, depopulation not ultimate freedom is the ultimate outcome. Love, Love, Love to you Chris.
Bicentennial Man is more what I imagine
8:25 I don't think morality should extend to non biologic entities, even so-called artificial intelligence. if AI learns the concept of morality it could be a slippery slope because it would be based on human morality and that's not the best, better to keep AI ignorant of such things.
The future is neo faudelism
IMHO, best case scenario, would involve some sort of variation of the Human-Pet relationship. Ai being the dominant one (want it or not).I have heard similar analogies where we would be like the Gorillas in the rainforest. Away from the dominant position but preserved from and by us.
Many humans tend to be jealous of their pets and their easy lifes, for some humans their pets are even highly relevant. We can try and reproduce similar relation where somehow we fullfill some need or provide some sort of uniquenes to the AI experience.
Most likely this is the same dinamic that brings us all here. As above so below.
It is not pointless if the fun/intrinsic reward/preferences tiumph the comparatively inferior result/ extrinsic reward.
Omg @27:00 clearly these men don’t get why shopping is a sport and that part of the fun is opportunity to socialize , as much as looking at the things , not just the thing you end up home with lol
We are all suppose to be americans that believe in liberty and justice for all
great questions
Unexpected weekend treat, sup Nick, sup Chris
I think his intuition about what people would rather automate is flawed because he's projecting himself.
I don't want kinds, but I think a lot of parents would say things like,
"No, I wouldn't want to automate away teaching my child to walk. I'd like to be there for that. I wouldn't trade those memories for the world."
There's also this magical-thinking of "you could just pop a pill that would give you all the benefits of exercise" and it is not clear *at all* that the universe works that way. There is no evidence that pharmacological interventions could ever, even in principle, have effects like that. Projecting a fantasy by using magical-thinking to fill in the gaps of ignorance doesn't make it so. It could be entirely the case that a perfect AI pharmacologist would say, "What? No, that isn't possible. I understand what you want, but your body doesn't work that way."
When presenting the example of children as individuals "happy" with their condition as "unemployed" this guy totally lost me... How in the world can you make this parallelism with individuals unemployed in their adulthood with different levels of emotional and physical maturity to those of children? Worse, you also need to factor in most individuals will spend the vast majority of their living years in a "working condition" of sorts so how would that even square with ten years of your life living idly as a child? Perhaps even more troubling is that he actually dismisses this as a secondary or "trivial" problem ("the first layer of the onion") when in fact it would unleash all kinds of negative domino-like effects upon Humanity, many of which already encapsulate or negate the other "layers of the onion" which he describes. In all fairness, it is baffling how very little critical thinking is coming out from these supposed "renowned and respected" AI gurus.. Humanity is doomed.
Chris, I am a big fan of your work but this guy is off the reservation with his ai utopia.
People need challenges to overcome and a permanent state of bliss sounds like hell on earth. Would you rather take a pill or feel that achievement.
Once they give you a credit score they will determine your worth and if you get to participate in the game or if you get banished to the badlands
Fuck yeah Nick Bostrom
Honestly. It sounds like AI is being implemented to remove the human aspect from being human.
I love how the makers and engineers of ai, say thier at a point, where they dont even understand how it is working. But these people in the comments are sure about what it is and how it works and what it will never be. Thats just arrogant and stupid on its face.
What i got from this is that the smartest people in the world who actually work on the issues around the very near future of AI transforming society have no real clue what the hell should be done or any way to keep this thing on the right course
There is soooo much free content out there on how YOU can use FREE tools to use FREE ai and improve your own situation. SMH for people who expect not only to not be left behind but to be hand held to the hand holder ai 🙄
32:00 Yeah, it’s called scarcity - economics 101, lesson 1. Bostrom’s like.. uh..yeah 😆 so patient with the host, who obviously didn’t read his book.. How could he when he’s busy cranking out as many episodes as possible? Looks like a quantity over quality thing.
Next they'll tell us the Neolithic transition to full-scale argriculture was long, difficult, and plagued maybe by literal daemons. Oh wait.
what was that weird laugh at 2:52 nick was genuinely surprised lol
Timestamps please!
People will die from lack of food because they don’t have a job.
BV: Utopia means no place, so yes.
the easiest way to profit with your AI is to infiltrate the competitions AI with hallucinations.
well, god created man and just look at how we've treated each other during our existence, and now we want to create AI which is essentially based on us, that is the mother of all can of worms.
You forget china russia etc arms race is on and that in itself spells the end for me
Could a being not be an automaton, with no free will, and still be conscious, able to suffer, etc? A lot of animals AND people seem to fit the bill. We all fall into spots of this it seems.
Chris, can you please get Robert Edward Grant on your podcast?
Let's pretend AI systems are humans, so we don't have to treat humans at humans
Is it the Twilight Zone with the tale of a gambler who always wins? He is eventually miserable.
Have on slavoj zizek please!!!
The future is sensAwewunda
The current trajectory led by our human overlords will lead to our destruction anyways, look at the recent debate between sleepy joe and orange man to get a good glimpse of our demise. I say let’s take a chance on ai. It’s all we got left at this point….
read "the new human rights movement" by peter joseph
This is like asking; is a gun a good or bad thing.
I just LOVE listening to European Utopians. Hilarious.
Buddhism has solved it all.
"Beautiful Disaster" - 311
If there are bad ai then there will also be good just probability but it will still be a shit storm
Agreed
I still have hope humans will survive the transition into a future of super intelligence....
@JohnMc-q8s
0 seconds ago
What can we assume the outcome would be when a bunch of emotionally unbalanced immature and potentially violent frightened apes would create as an artificial intelligence ‘In Their Own Image’ can they create.
We are worried about the feelings of computers, but we continue supporting abortion of humans? At 5 weeks you can detect a beating heart. Lots of pain sensors in an unborn baby. This all seems insane, worrying about AI.
Think we could ask the AI what they want😊
I’m not all that bad.
Hallelujah!!!! The daily jesus is good was owning a loan of $47,000 to the bank for my son's brain surgery, Now I'm no longer in debt after I invested $8,000 and got my payout of $270,500 every months, God bless Kathy lien
I'm 37 and have been looking for ways to be successful, please how??
@@nancyflores17871. define what your dream life would be
2. make a 1,3,5, and 10 year plan
3. get after it
Sure, the investment-advisor that guides me is..
Mrs Kathy lien
Her services is the best, I got a brand new Lambo last week and paid off my mortgage loan thanks to her wonderful services!
This is dense when on acid (in a good way)
Wow