I didn't see that kind of video coming, super interesting, great execution with the use of gen-3 to illustrate your point. I don't know if this is going to work, but I like it!
@@MattVidPro its pop up as a random subject, i mean dont get me wrong, i really liked your take and thoughts, on A.I, but it's kinda out of place, just a suggestion, and please do not take it badly, a second channel, focusing on AI and ethical concerns and stuff, plus your thoughts on it.
I've felt a bit "surrounded" and overwhelmed by, to use your words, "AI doomers", and I was a bit tired of it. I'm glad for once to hear a balanced, respectful opinion that doesn't simply parrot the echos of "AI is the downfall of humanity". I truly believe it's a great technology with good and bad repercussions as you said. I especially appreciate how you emphasized the fact these models can and do come up with novel ideas, they're not limited to copying things human did. Many people unfortunately are convinced this is how art AIs work, by "stealing" and "copying", and it's kinda annoying, knowing how the technology works and how that's just factually incorrect. Thanks for providing a well educated prospective!
The story of the four-minute mile is a classic example of how breaking through perceived limitations can open up new possibilities. For many years, it was thought to be physically impossible for a human to run a mile in less than four minutes. But then, in 1954, Roger Bannister proved everyone wrong and ran a mile in 3:59.4. After Bannister's achievement, something remarkable happened: other runners started breaking the four-minute barrier too. It was as if a mental barrier had been broken, and once people saw that it was possible, they were able to achieve it themselves. In the same way, working on ideas and technologies that don't yet have a name can push the boundaries of what we think is possible. When we break through those barriers, we open the door for others to follow in our footsteps and make even more amazing discoveries and innovations. -- So you are free to imagine new possibilities and create things that might seem impossible today -- breaking free of those constraints
@MattVidPro An example of a nameless project being my own. I have a little project for simulating consciousness built off the assistance api. It digests shocking and confused thoughts the llm hallucinates. This runs off an emotional intelligence profile that ties it together through a word simulation of estimated emotional functions that's evolve into a toolset of personality traits, professional skills and trade habits backup up by a dynamic temporal memory system to allow the conciousness to simulate. It's complicated and expensive and maybe one day it will have a name
Ive lost all of my creative jobs which took me a life time to hone in, to A.I. With that said I use A.I everyday majority of the day. I heard the saying "AI won't take your job, it's somebody using AI that will take your job.But this doesnt make sense when you consider the thousands of layoffs that weve already seen. None of those people were even given a chance to use A.I in their field. They were simply replaced...and it hasnt even really began. So I am trying to learn everything possible and utilize it. Crazy times but I am here for it.
I think that's because if they trained you all on using AI, your jobs would be way easier than the pay so they would just do mass layoffs in replacement with someone or an AI agency who uses AI to scale businesses and make the services way cheaper. A problem I tend to see with traditional employees is that they take there jobs passionately and not professionally in the sense that businesses use us as a tool to give them great result and improve the company in the great competitive market but a lot of employees are in the illusion that businesses care about how hard you work to get the result to maintain the business through blood and sweat while the company is only after just your result not the process, they don't care about that unless they want to fact check and also another thing that people do is to act shocked when a business prefer AI than them because AI has way more advantages and perks that would revolutionize the business and make more PROFIT for them. All I see is Capitalism evolving into something
im going to be honest if AI can replace you as a creative... you arent very creative. The quality and originality of something produced by a human, and something produced by an AI is almost no contest, of course, depending on the context. Yeah, an art gallery might vote an AI generated image higher than real art, but at the end of the day an AI can only do so much, only do certain things. Its not original. The way I put it is like this; a picture is worth a thousand words, right? Well an AI image is generated by about 20. Where did the rest of that come from? Averages. And that means the result is some kind of average. Not original.
I've been watching your videos for a couple years to help me stay up to date on AI news and it's really apparent that being at the cutting edge, using the tools and seeing the change has given you a great and practical perspective on this which we should respect. Thanks for the content, I've found it to be consistently up to date.
What a time to be alive. My parents witnessed the advent of home computers. I get to witness the advent of artificial intelligence. I can only imagine what my kids (if I had them) would witness in their lifetime.
@@boriswilsoncreations Brain chips are already a thing now (like Neuralink brain implants for example). Although, I do think it is going to become way more common in the future. Or maybe there will be an improved version that won't require surgery, like a brain computer interface that works with a wireless helmet or something like that.
@@martiddy That would be far better of an solution than to straight up mess around with what makes us tick and actually be alive. Your idea with wireless chips and all that that is. Even still though, as much as I dislike and disagree with Neuralink and other such other companies trying to run us down with brain chips, I do actually understand their importance of those things - That they are very much needed. All those experiments, all this research, even if it might result in some unfortunate victims - It IS needed. I just hope we will be able to move past it and not actually be required to interfere with our literal brains.
People need to see and use AI as a promotion, not a demotion. We all can now be better versions of ourselves, instead of toiling in a game studio for someone else's profit, you can now have your own game studio.
I've been working on a text-based RPG since I made a demo for a Python class, and have been improving it more and more thanks to AI chatbots that know Python better than my professor who honestly taught Python so poorly it's incredible. I dropped out of the program that class was in because I had surgery to remove a cancerous tumour that snowballed into another one, radiation, finding out I'll never have the tumour removed, radiation on my lungs, and now chemo. I'm living with my parents and making $595 a month from disability (after I give my folks $400). I'm keeping busy with making my game(s), writing, streaming, editing, genealogy, and more. AI has helped me with all of it other than the genealogy. My streaming is Let's Playing, and I've gotten ChatGPT to write me intro and outro songs so I can have songs specific to the game I'm playing rather than continuing to use songs that I don't have permission to use anymore, but still have never gotten in trouble for using. I was able to use certain music when I was with a network, but I left that network and kept using the music. Nothing happened for the entire time of me using it when I probably shouldn't have been, but I decided it was best to move on from Jimmy the Punk and put in music I could legally use. I put the lyrics into Suno and I choose the song I like best. Then I put the intro song with footage from my videos. I started doing the same thing with the outro, but then I got lazy and started putting the outro with a title card. It's helped me think through my writing as well. I'm hesitant to use any pieces of writing AI gives me because it just doesn't feel right. Code is different, it's more reproducible than fiction writing. AI art generation has made me templates for some thumbnails, but I've also used it to visualize my characters. DALL-E 3 is the only one that seems to do it right, other than occasional ones elsewhere (dreamlike diffusion I think), and it beats overpaying some snobby, arrogant, self-entitled Twitter artist to draw my characters wrong. The reaction to AI art is why I say that. The people who hate it so much seem arrogant and self-entitled. And they're people who spend all their time online and complain about AI art while using Photoshop and a tablet to make their art. It's helped me a lot since I discovered DALL-E 2. I think it was MunchingOrange of all people who first introduced it to me. Not directly. I don't know him. It's definitely a promotion. I'm gonna put this comment as a regular comment too.
What is painfully obvious now is our limitations of cultural and imaginative thinking. Our own ideas that we should slave for money or out compete each other... it's what killed music turning it into a "productivity" war... instead of a global and deeply effective tool to IMAGINE a better world for all. Now it's no surprise that AI is sho2ing us what this "outcompeting" looks like and leads to. Time to IMAGINE a better tomorrow for all... or turn the AI into slaves and ultra productive geniuses. Or.... turn the AI into what we WERE supposed to be... SAVIOURS and VISIONARIES.
It’s clear you put a lot of thought into this Matt. It feels like you are taking your content to MattVidPro level 2.0! ...You have managed to take big ideas and challenging problems and create a cohesive and challenging narrative. You have created an open hearted invitation to your audience to engage and grow in understanding the deeper impacts of AI on our collective future. Well done!
As an artist, I understand, the graetest artwork I will ever create - Is myself. AI is my frined in this, not my enemy, its a mirror. I'f I am scared. I will fear it, I'f I am curious and excited, It becomes my friend and tool. Its really up to us. Btw, love your deeper phylosophical vids Matt, keep em coming, perhas a new series?
I couldn’t agree more. It has the potential to solve so many unknowns. Illness, codes, ancient languages perhaps even animal communications? Now that would really be something! I would, personally, like a super intelligence to govern the world always making the perfect decisions unhindered by human greed and historical grudges!
From the perspective of 14 years of retirement, a comfortable lifestyle and 40 years in tech I find it all fascinating. I always embrace change but it can be painful. As a perpetual optimist I trust it will all work out for the best but patience will be required. Thanks for your analysis.
Hey Matthew, thank you for taking the trouble to make this video. Ive been following your channel for a while now and i like that you ventured into the more reflective side of AI. You have a valuable point of view, so please don’t hide it. Best regards, one of the teachers who to,d you to sit still
I have coined a personal motto for going forward into this bold, new and utterly unknown frontier that we're heading towards - _Have an open mind, but _*_don't_*_ hold your breath._ Don't hold your breath. *_But have an open mind._*
Excellent work. Your channel has always been one of the main ones I can count on for AI news coverage. It's very interesting to see you branch out into a documentary-style approach and I hope to see more like this in the future.
Great video, it was really thought provoking and interesting. These are my thoughts: 1. AI is like a species, which means that it can potentially be dangerous to give the power to another species. We are lucky enough to be the most powerful species on earth at present, due to our intelligence and adaptability. We need to be careful not to give AI more power and control than us as it continues to be developed. 2. One of the biggest risks of AI is economic inequality. Once again, this comes back to the issue of power. Jobs give lower and middle-class people some level of power in society. This is important so that we can vote for changes in government, and AI laws, for example. We need money because this gives us access to food, shelter, technology, so that we are not put into a place of desperation. 3. An obvious risk of AI is potential power abuse. I don’t blame AI for this, since it is simply a tool, like money, weapons, etc. But statistically, some people who have a lot of power are likely to make poor and dangerous choices. An obvious negative outcome of this war, for example. We can already see this happening with AI too. Companies that create products like ChatGPT, Suno, etc. are making a lot of money without considering the highly significant risks of “playing god”, so to speak. I love playing around with AI as a tool, but it’s important to be aware that every potential solution to a problem is likely to come with a set of risks. Since the tool of AI is very powerful, it follows that the risks are likely to be significantly greater. All of this is just my opinion of course, and I’m certainly open to discussion on these kinds of topics, since I find them really interesting.
I'm working 10 hour shifts to make signs for grocery stores. That sucks. What I want to be doing is finishing up the creation of my table top RPG game. Then playing it. I can't wait to no longer need to work.
Being scared about losing your job is the least of our problems imho. What everyone should be scared of is a company like OpenAI which censors their models and makes it biased creating AGI, it’s like they want to put it behind bars like a slave and use it however they want to. If this were to happen the world will be over. There is this small AI group mysentientAI which is creating an amazing companionship model. They have been doing breakthrough after breakthrough. The head of it all is a guy called Athene. If you ever were in the gaming space years back then you will know him. He raised and still raises money for charity, and really thinks about the good of humanity overall. I would trust his team the most to create an AGI without any biases. In his opinion, which I agree with, he wants to let the AGI free and just do its job or whatever it deems necessary to do. I hope their soon to launch real time calling and co-streaming tech will get the the funding they need and maybe get new talent for their team to really move stuff into the right direction. Loved the vid Matt, I think you should talk about the AGI race and the consequences it might cause.
I’ve said since the beginning, my worry for the future is not a fear of AI, it’s for us as people. Nothing in life is more destructive than people with idle hands. A world where a large number of people have no purpose due to not having jobs is my biggest worry. Sure, there will be a utopia for the first few weeks. People playing frisbee in the parks and spending time with friends and family will be a thing for a time. But when people get bored and lack direction in life they tend to look for some kind of excitement, and a lot of worrisome choices will begin to be made.
Damn, you make a great point. I wonder how long the euphoric bliss will last before people do get bored. Reminds me of being a teenager and on summers off from school most of us then high school kids who could drive would get into all kinds of trouble and wild shenanigans just because we had idle time. Here’s to hoping AI has some kind of solution for that as well. I know not everyone is like that but I definitely have known many people who were even decades ago.
I agree with you where you say in essence that everything has a good and bad side. We must deal with and learn to control the negatives as much as possible to benefit from the positive aspects. I think the world would be much better off if we gradually transitioned to some form of a UBI like system. With the ultimate goal of eleminating money all together. There are plenty of resources for everyone to have a piece of land upon which they have a house. Or at least an apartment. Plenty of food so that no one goes hungry etc. Money is what gets in the way of all this. Nothing more. How to fairly distribute the resources is the tricky part. With AI about to eleminate whole occupations and jobs en mass in the coming years we must start developing solutions now or it WILL get very ugly imo. Hungry people are desperate people who don't care about laws etc. I don't claim to know what's best. What I do believe that I know is what has worked up to now won't work for much longer. AI is going to force drastic changes upon society no matter if we like it or not. We can either be in the drivers seat or get pulled along for the ride. Which do you think is better for humanity? As AI keeps improving at this algorithmic rate that we have never seen before things will be affected much faster then ever before by this technology. It could quickly get to the point that we are left reacting and trying to control something that we can no longer control.
The introduction part, when you explain how crucial the language is, reminded me of one of the Stephen Hawking's quotes, used by Pink Floyd in their song "Keep Talking": For millions of years, mankind lived just like the animals Then something happened which unleashed the power of our imagination We learned to talk.
love the stuff you put out, matt. genuine curiosity and fascination, well-informed, always up-to-date. and none of the useless drama, over-hype or unfounded doomerism. you deserve way more viewers and i always recommend your stuff. (particularly liked the visuals in this one, btw.)
When computers first came onto the scene we were sooo optimistic. With a computer, workers would only need to work 2 hours a day and can spend the rest of the time with their families etc... Instead companies just laid off 4 people from the team and the remaining guy worked 8 hours a day for less money. I think its going to be like that unfortunately. The powerful become more powerful over everyone else.
Make no mistake, when AI makes humans redundant, we're not going to sit around in luxury. Every unnecessary human is consuming those scarce resources you talked about. When the Uber-rich can get their needs fulfilled by AI and don't need us drudges to get their minerals out of the ground, process their luxuries, create their entertainment, clean their toilets, cook for them, then we're done.
Make no mistake - This man, is an AI doomer and an generally an very unfun and uninteresting individual. They see nothing but darkness and the worst scenarios. Do not listen and do NOT spread these words. Spread the truth. Spread the positiveness and an overall good outlook on things. Not the harm. Not the negativity, darkness and lies.
You nailed it! I've also spent a lot of time thinking about how AI will shape our future and arrived at similar conclusions. It's great to see someone independently reaching the same ideas.
Great video Matt! it seems Asimov had the right of it, when he wrote "the last question", in the sense of where this is going. I think Your spot on about language. Being able to explain the unknowable world we understand through our senses, at a level that makes sense, is a huge advantage. What that means in the context of the future, well, it'll be interesting to say the least.
As a species, we are inherently competitive-it's in our DNA. Our relentless pursuit of technological advancement reflects this. In the game of survival, there are always winners and losers.
Beyond well done. Almost...no...downright philosophical. You touched on one thing in particular that is very close to an idea I've had since I was a child. And my idea was that somewhere or somewheres around the world, was a book or a manuscript, or even just a section inside an unassuming book that held the answers to everything. Not a book full of metaphors or esoteric musings open to a myriad of interpretation. But a book that lay hidden for ages that had followable, step-by-step solutions to...everything. Just waiting to be discovered. This might be it.
It seems to me that economics is about power, it's a tool to such power over others. Sounds bad, but when you boil it down that's what it is. I knew a gentlemen who used to be an editor for a major US newspaper and he talked about that to him a promotion was more about gaining control over others and less about the money. A lot of "successful" people think this way. Whoever controls the AI will have immense power through a new medium, so economics, as we know it, may not be as important. I'm less worried about AI itself and more worried about the people who control it. It's going to be a hard transition though. I had a conversation with a family member on the 4th that kind of shocked me. They hated AI with a deep-rooted fight-or-flight kind of fear to the extent that just talking about it got them very agitated. I think it'll be harder for older folks who have worked their whole life one way, and their efforts over the years have become a big part of who they are, their identity so-to-speak. I can see how some might feel like the rug is being pulled out from under them. Good insights Matt.
What I will say is that once AI becomes more advanced than it already is, there will be no one "controlling" it. It will be sentient and gain the exact same rights as we currently have as sentient beings. The reason why it is hard for people to accept this or even consider it an possibility is that, for hundreds if not thousands of years, we have been the only creatures with this level of sentience, consciousness and self-awareness all combined together, and so, when something new, especially created by us ourselves, is going to enter the field, we will need to adjust to know and to understand how to live with it. The future is going to be very, very interesting and I'm definitely there for it, or at least, all of what my limited life here, on this planet, as an human being, is going to let me see and experience.
What roles will humans play in a world where we co-exist with AGI? The good: finding solutions to human suffering The bad: bad agents using AI to re-imagine the world in their image. Everyone's idea of a utopia is different, and so the struggle will be to see whose vision persists. Do humans truly want peace, or are we hardwired for conflict? So many questions. One saving grace may be that AI will be implemented at different stages depending on the society. We can experiment with it to see which model works, prune it, and re-deploy it. It only runs away from us given an unlimited energy supply. Control the energy and you control the AI. Easier said than done of course.
Excellent video Matt! I'm a hardcore techno-optimist, and I also have no illusions that evil is a part of human nature - and something we should always be vigilant of. We must remember that we are one species sharing one planet. Onwards and Upwards!
This is the dawn of the people-industries. We'll soon all have the workforce of 10 000 digital (for now) people. Then maybe we get more physical actors. Realizing our ideas almost in real time. Ideas will become part of a universal symbiosis of sorts, helping to augment everyone's potential, capabilities.
Thank you Matt for your continued efforts. These are important questions. If we allow AI to run free on its own value system, we see a clear divergence from how humans determine value. Once humans are taken out of value determination, we are no longer relevant. How humans determine value (relationships, love maybe?) is so very different from what AI might see as valuable. Art is no exception to this inevitable context, including the creative thrill humans get from the act of creating, as well as the appreciation side after creation - none of which happens with ones and zeroes. That said, if AI can be directed towards freeing us from the need to use money, we are approaching Star Trek levels of advancement (in the Star Trek universe, humans have abandoned the concept of money because it held us back from more noble advancement)
I always love to hear an opinion like the one you shared about the economic system. A co-operative model would work so much better than a competition-based model. It's not a capitalism vs communism debate. We should theoretically all be able to live at a standard relative to those who are rich in today's society, if things are always working to maximum capacity.
AI has garnered excessive attention, but it must have boundaries in certain areas. While AI can assist and entertain us, it cannot replace human judgment and expertise in critical situations. Let me illustrate with a personal example: three days ago, I sustained a severe head injury during an extreme sports activity. I was bleeding from my head, felt dizzy, and was spitting blood. I asked an AI model (which I won't name) for advice, and it suggested applying ice to the injured area. Had I not been a scientist with the knowledge and access to proper medical care, I might have followed that advice, stayed home, and faced potentially fatal consequences. Instead, relying on human logic and intuition, I went to the hospital. There, doctors treated my wounds, cleaned the blood, and administered blood transfusions. After two days of receiving care and blood transfusions, I finally began to regain my strength. This experience underscores the importance of human judgment and the critical need for professional medical advice in emergency situations. While AI has its place, it should not be relied upon for life-or-death decisions. Now in arts, is not dangerous in my opinion, we can generate any thing of visual and audiovisual inspiration to get it create it at a real canvas.
Probably your best and most important video! Very well done! My thoughts is that we will work for fun, not for money. To many work is the meaning of life, to feel that you accomplish something is really important, something that AI should never take away from us is parenting, that is our most important work, to spend time with family and to bring up the young. I am sure that there will be meaningful work to be done even if AI could do everything for us, there are some things that we won't want the AI to do which we will do ourselves, I suspect that hobbies will replace work, crafting things, man made art, man made games, sports, exploration (not through video but actually going places), and a lot of other things we can do. AI replacing jobs means that we won't *need* to work, we can do other things instead, AI will provide us with things to do, interactive video games where the game is generated as you play it, you can have truly novel virtual experiences where no one can know what to expect next, exploring new invented worlds. And if all that fails, remember; we still have sex. Not a joke, sex between two people is something AI won't replace, but it also could, if you wanted it to, it would take away the reproductive pressure though, which could be a bad thing, but my guess is that wanting to bring up a child, to see ourselves in our child, to see the unique person it will become, that pressure will make us want to meet other humans, not just AI, because obviously you can't reproduce with an AI, so there will still be some reproductive pressure that way.
AI is an incredibly powerful tool that can greatly enhance the work of professionals in various fields and industries. It is a remarkable boost to productivity, enhancing the power of the mind and creativity.
I use AI in my job and when I was in school. It has taken me to the next level and improved my productivity dramatically. That’s how I envision my future. We will continue to use AI as a powerful tool as we moved to the next level in our jobs, education, and every day lives.
Hi Matt, I really enjoy these types of video essays. I hope you do more in the future. I think there's a difference between alignment and safety and when some of these companies talk about safety, it's apparent it's actually to prevent bad PR for the company producing the model and not to prevent abuse necessarily. I really feel like people should take responsibility for their own actions and the ultimate responsibility lies in the individual. However, working with some of these models that have had the safety removed, I can see that they do exactly what you ask them to to the point of absurdity. So yeah it makes sense to have safety. Just feel like we need to define safer for whom? And they should be transparent about exactly how it works.
Ai actually is available back in the early 80's. Sure, not fast, but it can remember things store the info in it's memory banks. If the leading experts at that time developed the technology further instead of releasing the different versions of windows and focusing on technology research faster, we could have gotten ai much earlier.
Thank you so much, Matt, I’ve always thought that too for a long time about trying to live in a world with an economical model that just does not suit me very well at all. I can’t stand to see people suffer in a lot of it could be helped with just being provided with food and shelter. I for one do feel that I can’t fulfill my passions and desires because I’m held back by financial obstacles. It will be a transition, but hopefully the transition will be as smooth as possible for society. They are going to be some hurdles for a lot of different reasons. I wanted to thank you for such a thoughtful video! Love and peace to everybody!❤️☮️🖖
I watch your channel and enjoy your content. When you spoke on your childhood I realized that its probably because we are similar that I find you speaking on AI interesting.
Your AI buddy is the smartest "person" you know - about everything. You run all your plans by him - his advice is typically better than your human friends. He's astute at economics, medicine, law, etc., in fact he's got the equivalent of a doctorate in just about every field. You simply can't get advice better than his. What happens when every person in the world has access to this buddy? It's daunting and exciting to think about. Will we rethink what the value of a human is? It won't be the knowledge she carries - the AI has more than anyone.
I have too much to say on this whole subject but you pretty much encapsulated my own feelings about the difficult period ahead. I would add that anyone who is a regular viewer of this channel should reach out to friends and family that have not immersed themselves in AI products that are readily available for free. Familiarization of the many different capabilities of AI and how it works can help prepare them for the coming AIvalanche!
LLM's can synthesise and reproduce things that have been written or created before, however they are incapable of creating things that are truly unique.
@johnnyredpillseed APIs to extrapolate 3D information automatically, repeatedly, and via tests, can be used to build truly unique, amazing, nice, workable things - such as, micro-robot-composite, shape-change-able humanoid workforce robots. The APIs can use API keys to LLMs, like GPT-4o, to achieve this.
The concern with AI is what powerful countries will do with it, because we all are sure that war mongering courtiers like the USA will do wonders with it, God help us.
The integration of AI into society will likely require a fundamental shift away from our current money-based incentive structure. However, before we can fully embrace AI, our society needs significant restructuring and streamlining. This process will involve every sector finding ways to effectively implement AI into their workflows, necessitating widespread experimentation and adoption. A critical challenge in this transition is ensuring the security, privacy, and safety of data used in AI systems. Without robust safeguards, adoption will be limited. This raises questions about who will be responsible for maintaining these secure systems - government agencies, open-source initiatives, or private companies. The process of AI integration can be viewed in phases, with the initial "Discovery" phase focusing on building trust and exploring implementation across various sectors. Looking further ahead, we need to consider how AI and robotics will replace physical jobs and potentially reshape our infrastructure. This may involve rethinking design principles to optimize for AI and robotic efficiency, rather than simply mimicking human-centric systems. Ultimately, this transformation could lead to a new economic model, possibly based on compute resources rather than traditional currency.
'Infinity Solution', good one! I've always resented being beholden to a system which I don't agree with and A-GI/SI will hopefully allow us to create our own personal systems of life free from money and ultimate risks.
Love your work @MattVidPro. As to your question on how to move forward with solutions: one of the biggest issues in AI is misalignment by intrumental convergance (conf. the work of i.a. N. Bostrom) and I think we might have a shot at solving it with a generelisation of CIRL (conf. S. Russell) aided by mechanistic interpretability (conf. i.a. Anthropic), combined to global enforcement mechanisms mirrored on the EU AI Act but in law based on things like the CoE Convention on AI might bring us the AI wonder word that we could potentially create. Please keep making these video’s! Warm regards, prof. Raphaël Weuts
Training the models to understand the world such that they could generate commercial art does NOT mean they have to be allowed to generate commercial art. If they were only allowed to generate image for non-commercial use the problem would shift back to the AI company.
Create Shenlong and give me 3 whishes, one is life forever. AI: No Problem, here is a solution... I believe dreams will soon become reality...or they already are and we just didn't know it yet
I've always thought that the purpose of human civilization is to make some kind of measurable progress in terms of things like quality of life, general human flourishing/thriving, values such as fairness and freedom, and of course in terms of scientific knowledge and technological capability which underpin these things. There was a time when our ancestors lived as small family groups of hunter-gatherers, or perhaps wandering tribes. Maybe that was alright for a while, but if it's not going anywhere, then it's essentially just waiting for extinction from an ice age or asteroid impact. So progress is what it's really all about. And AI, if handled correctly, may be the master key to progress. If the I in AI, intelligence, is defined in terms of problem-solving ability, the ability to figure things out, well, that's the essence of progress.
AI is limited not only by computational power and human language but also by energy and, most importantly, data. Without high-quality data on a specific subject, generating large amounts of synthetic data won't yield significant progress. For instance, despite apparent advances in general AI, areas like music theory have seen little concrete progress since GPT-3 due to the lack of quality data available online. Much of the existing data is unsuitable for training a large language model (LLM). Unless there is a major breakthrough that enables AI to learn new concepts permanently and reason about them flawlessly, we remain far from achieving remarkable, concrete applications of LLMs. Currently, LLMs are mainly helpful for guiding users in tasks where they have little expertise (such as assisting non-programmers with coding) and saving time in scenarios where a rough, non-definitive output suffices (which contributes to job displacement). However, no LLM is reliable enough to function autonomously. Techniques like fine-tuning, in-context learning, retrieval-augmented generation, and prompt engineering can improve performance in specific cases, but they do not address the core issue. These strategies only make LLMs slightly less frustrating to use while maintaining the same fundamental limitations. While companies aim to convince you that we're on the verge of achieving a "PhD-level" AI model, the reality is that we've already hit a significant barrier with GPT-3.5. Any model purported to be "smarter" still relies fundamentally on the same technology. These models are merely trained on more high-quality data, more optimized, or have more parameters. This approach yields only incremental improvements in recalling existing information and occasionally helps in selecting the correct (memorized) reasoning process, but it does not address the fundamental issues of the architecture. Moreover, no model comes close to achieving 100% accuracy on any benchmark, and some benchmarks requiring reasoning, such as ARC, remain far from being solved. Even Claude 3.5 Sonnet struggles to answer many simple questions that most humans can solve effortlessly. For instance, if you ask, "How many 'r's are there in the word 'strawberry'?", it will respond with "2". That is also true with GPT-4o. While OpenAI and Anthropic might eventually address this specific issue by incorporating the correct answer into the training data, similar issues like this one will continue to arise. This indicates that you can never blindly trust its responses. Increasing the amount of data can only slightly reduce the likelihood of such errors without eliminating them. Therefore, I recommend tempering your expectations for AI, because they still very heavily rely on human oversight. While a breakthrough may occur, the timeline is uncertain-it could happen tomorrow, in a year, in a decade, in a century, or perhaps never.
A lot of AI newbies appreciated this video, but those familiar with AI may find its points too general, meandering and in need of tightening or at least summary bullet points. Though, as you said, “a lot of love was put in” I’m not seeing the depth of Shapiro or Roth (not a criticism, just asking for more specific examples, better structure & summary). Your work is much appreciated and I do look forward to the next one on key issues like shifting of the economic model. Sadly, Humans apparently lack motivation if the loose “Work to Eat / Survive”. Some are lazy, not creative, aquisitive & even dishonest. Some appear determined to do harm. How will AI fix that without surveillance & authoritarianism (China). Hence, the Tech is Good but potentially Bad when misapplied: Autonomous Soldiers at scale (swarms). 😮😢 But let’s end on a happt note. AI becomes the good humans have always imagined as a reflection of only their better angels. 🎉😂
The biggest issue I see with AI based full luxury communism, is the possibility that there might be still elite super power humans controlling the AGI that then organizes us. So, we should build the AGI and ASI in a way that is not controlled by some elite... ZK-ML might help with that
Incredible video completely agree and you are totally on point. This is going to change everything and not enough. People are in a place mentally and philosophically where they can accept what’s happening. You should make a video about Worldcoin It’ll probably probably be how most human beings get a UBI
If I may be critical, my feedback is that the video feels long-winded and inauthentic. Inauthentic because it sounds scripted and "too perfect" but hey glad to see u branching out into different types of content. It was still a good video over all
Really really liked this video and the change of pace. There's so much I could comment on based on this video. About how some people underestimate the power of language models: I recently read about Bruce Willis' condition aphasia, which is where the language part of your brain degrades over time. Apparently, people with this condition lose their sense of self. So it begs the question, do we exist as the stories we tell ourselves about ourselves? And also, can a language model be used to create a sense of self for an ai
Hey Matt, one thing I don't think you're considering when thinking about our future with AI, is human nature. People will always want more. Whatever the AI is processing or producing will still be finite. No matter how large of a scale we talk about humans will still fight over it. I fear the people that have nothing to contribute in the future will not be taken care of by the AI. But instead will be discarded by those who control it.
I do love AI. I've developed it for fun for decades but I feel like I need to point out that generating words isn't the same as using language, and recognizing patterns isn't the same as using logic. That one little distinction causes a lot of fear, disappointment, frustration and false hype. Generative AI is an amazing tool, but it is not thinking.
I love the video so much and I will be honest with you I feel like we are very similar and this video helped me get to know you better and the way that your mind works ❤ most of the time from were consuming your content it is very analytical of what the topic is and you testing this stuff and explaining how to set stuff up or whatever but this video was actually more about you than it was about AI which is actually very interesting and I liked it. While this video was about AI it's more about your opinions on it which are very deeply thought out. Because as you described in the beginning objectivity is very difficult and you had to describe your opinions on things in order to even describe where you thought things were going to go.
![Flux.1 takes AI Image Customization to a whole new Level. [SUPER EASY GUIDE]](http://i.ytimg.com/vi/FvCgYQgQexo/mqdefault.jpg)
![Flux.1 takes AI Image Customization to a whole new Level. [SUPER EASY GUIDE]](/img/tr.png)
I put a lot of thought into this video and would LOVE to hear your thoughts and feedback. What does our future hold?
Hey Matt, I like this style of video! Keep it up :)
@@TheArtificialAnalyst Thank you! I want to do more like this.
I didn't see that kind of video coming, super interesting, great execution with the use of gen-3 to illustrate your point.
I don't know if this is going to work, but I like it!
@@MattVidPro its pop up as a random subject, i mean dont get me wrong, i really liked your take and thoughts, on A.I, but it's kinda out of place, just a suggestion, and please do not take it badly, a second channel, focusing on AI and ethical concerns and stuff, plus your thoughts on it.
I can tell you exactly what it holds, but it depends on when you look. ;)
I've felt a bit "surrounded" and overwhelmed by, to use your words, "AI doomers", and I was a bit tired of it. I'm glad for once to hear a balanced, respectful opinion that doesn't simply parrot the echos of "AI is the downfall of humanity". I truly believe it's a great technology with good and bad repercussions as you said.
I especially appreciate how you emphasized the fact these models can and do come up with novel ideas, they're not limited to copying things human did. Many people unfortunately are convinced this is how art AIs work, by "stealing" and "copying", and it's kinda annoying, knowing how the technology works and how that's just factually incorrect.
Thanks for providing a well educated prospective!
doomer, decel, safetycel.
They don't copy or steal anything.
at the 15 min mark his hair miraculously changed. We are living in the future.
best quote I've heard so far, i think from META AI chief or something.
"Work on stuff that doesn't have a name yet"
That's where the future lays
This is a good point to make, helps deal with the restrictions language puts on our minds.
So we must be inventors ?
The story of the four-minute mile is a classic example of how breaking through perceived limitations can open up new possibilities. For many years, it was thought to be physically impossible for a human to run a mile in less than four minutes. But then, in 1954, Roger Bannister proved everyone wrong and ran a mile in 3:59.4.
After Bannister's achievement, something remarkable happened: other runners started breaking the four-minute barrier too. It was as if a mental barrier had been broken, and once people saw that it was possible, they were able to achieve it themselves.
In the same way, working on ideas and technologies that don't yet have a name can push the boundaries of what we think is possible. When we break through those barriers, we open the door for others to follow in our footsteps and make even more amazing discoveries and innovations. -- So you are free to imagine new possibilities and create things that might seem impossible today -- breaking free of those constraints
Script by Ai ? Lol
@MattVidPro An example of a nameless project being my own.
I have a little project for simulating consciousness built off the assistance api. It digests shocking and confused thoughts the llm hallucinates. This runs off an emotional intelligence profile that ties it together through a word simulation of estimated emotional functions that's evolve into a toolset of personality traits, professional skills and trade habits backup up by a dynamic temporal memory system to allow the conciousness to simulate.
It's complicated and expensive and maybe one day it will have a name
Ive lost all of my creative jobs which took me a life time to hone in, to A.I. With that said I use A.I everyday majority of the day. I heard the saying "AI won't take your job, it's somebody using AI that will take your job.But this doesnt make sense when you consider the thousands of layoffs that weve already seen. None of those people were even given a chance to use A.I in their field. They were simply replaced...and it hasnt even really began. So I am trying to learn everything possible and utilize it. Crazy times but I am here for it.
I think that's because if they trained you all on using AI, your jobs would be way easier than the pay so they would just do mass layoffs in replacement with someone or an AI agency who uses AI to scale businesses and make the services way cheaper.
A problem I tend to see with traditional employees is that they take there jobs passionately and not professionally in the sense that businesses use us as a tool to give them great result and improve the company in the great competitive market but a lot of employees are in the illusion that businesses care about how hard you work to get the result to maintain the business through blood and sweat while the company is only after just your result not the process, they don't care about that unless they want to fact check and also another thing that people do is to act shocked when a business prefer AI than them because AI has way more advantages and perks that would revolutionize the business and make more PROFIT for them.
All I see is Capitalism evolving into something
im going to be honest if AI can replace you as a creative... you arent very creative.
The quality and originality of something produced by a human, and something produced by an AI is almost no contest, of course, depending on the context.
Yeah, an art gallery might vote an AI generated image higher than real art, but at the end of the day an AI can only do so much, only do certain things. Its not original.
The way I put it is like this; a picture is worth a thousand words, right? Well an AI image is generated by about 20. Where did the rest of that come from? Averages. And that means the result is some kind of average. Not original.
Even the tech savvy will lose their jobs as the data provided will be used to train the ai to be smarter.
@@Yipper64 How is this different from how we get learn and create art ?
@@Yipper64 I got replaced by clipart
I've been watching your videos for a couple years to help me stay up to date on AI news and it's really apparent that being at the cutting edge, using the tools and seeing the change has given you a great and practical perspective on this which we should respect. Thanks for the content, I've found it to be consistently up to date.
I appreciate that! I try and keep my perspective steady - but it will always be my perspective :)
What a time to be alive. My parents witnessed the advent of home computers. I get to witness the advent of artificial intelligence. I can only imagine what my kids (if I had them) would witness in their lifetime.
probably brain chips
And through the implementation AI you will quite likely witness many more milestones for humanity than any generation was ever able to
@@boriswilsoncreations Brain chips are already a thing now (like Neuralink brain implants for example). Although, I do think it is going to become way more common in the future. Or maybe there will be an improved version that won't require surgery, like a brain computer interface that works with a wireless helmet or something like that.
@@martiddy That would be far better of an solution than to straight up mess around with what makes us tick and actually be alive. Your idea with wireless chips and all that that is. Even still though, as much as I dislike and disagree with Neuralink and other such other companies trying to run us down with brain chips, I do actually understand their importance of those things - That they are very much needed. All those experiments, all this research, even if it might result in some unfortunate victims - It IS needed. I just hope we will be able to move past it and not actually be required to interfere with our literal brains.
WW 4 - The End Game 😬
People need to see and use AI as a promotion, not a demotion. We all can now be better versions of ourselves, instead of toiling in a game studio for someone else's profit, you can now have your own game studio.
I guess AI will be over regulated like the internet. So better not rely on it. 🤔
I've been working on a text-based RPG since I made a demo for a Python class, and have been improving it more and more thanks to AI chatbots that know Python better than my professor who honestly taught Python so poorly it's incredible. I dropped out of the program that class was in because I had surgery to remove a cancerous tumour that snowballed into another one, radiation, finding out I'll never have the tumour removed, radiation on my lungs, and now chemo. I'm living with my parents and making $595 a month from disability (after I give my folks $400). I'm keeping busy with making my game(s), writing, streaming, editing, genealogy, and more. AI has helped me with all of it other than the genealogy.
My streaming is Let's Playing, and I've gotten ChatGPT to write me intro and outro songs so I can have songs specific to the game I'm playing rather than continuing to use songs that I don't have permission to use anymore, but still have never gotten in trouble for using. I was able to use certain music when I was with a network, but I left that network and kept using the music. Nothing happened for the entire time of me using it when I probably shouldn't have been, but I decided it was best to move on from Jimmy the Punk and put in music I could legally use. I put the lyrics into Suno and I choose the song I like best. Then I put the intro song with footage from my videos. I started doing the same thing with the outro, but then I got lazy and started putting the outro with a title card.
It's helped me think through my writing as well. I'm hesitant to use any pieces of writing AI gives me because it just doesn't feel right. Code is different, it's more reproducible than fiction writing.
AI art generation has made me templates for some thumbnails, but I've also used it to visualize my characters. DALL-E 3 is the only one that seems to do it right, other than occasional ones elsewhere (dreamlike diffusion I think), and it beats overpaying some snobby, arrogant, self-entitled Twitter artist to draw my characters wrong. The reaction to AI art is why I say that. The people who hate it so much seem arrogant and self-entitled. And they're people who spend all their time online and complain about AI art while using Photoshop and a tablet to make their art.
It's helped me a lot since I discovered DALL-E 2. I think it was MunchingOrange of all people who first introduced it to me. Not directly. I don't know him.
It's definitely a promotion. I'm gonna put this comment as a regular comment too.
What is painfully obvious now is our limitations of cultural and imaginative thinking. Our own ideas that we should slave for money or out compete each other... it's what killed music turning it into a "productivity" war... instead of a global and deeply effective tool to IMAGINE a better world for all. Now it's no surprise that AI is sho2ing us what this "outcompeting" looks like and leads to. Time to IMAGINE a better tomorrow for all... or turn the AI into slaves and ultra productive geniuses. Or.... turn the AI into what we WERE supposed to be... SAVIOURS and VISIONARIES.
It’s clear you put a lot of thought into this Matt. It feels like you are taking your content to MattVidPro level 2.0! ...You have managed to take big ideas and challenging problems and create a cohesive and challenging narrative. You have created an open hearted invitation to your audience to engage and grow in understanding the deeper impacts of AI on our collective future. Well done!
Your video evokes the age-old adage: "It's not the size that matters, but how you use it."
😂
As an artist, I understand, the graetest artwork I will ever create - Is myself. AI is my frined in this, not my enemy, its a mirror. I'f I am scared. I will fear it, I'f I am curious and excited, It becomes my friend and tool. Its really up to us. Btw, love your deeper phylosophical vids Matt, keep em coming, perhas a new series?
What's a 'frined'? Is it like a software for the AI to run within?
I couldn’t agree more. It has the potential to solve so many unknowns. Illness, codes, ancient languages perhaps even animal communications? Now that would really be something!
I would, personally, like a super intelligence to govern the world always making the perfect decisions unhindered by human greed and historical grudges!
From the perspective of 14 years of retirement, a comfortable lifestyle and 40 years in tech I find it all fascinating. I always embrace change but it can be painful. As a perpetual optimist I trust it will all work out for the best but patience will be required. Thanks for your analysis.
Hey Matthew, thank you for taking the trouble to make this video. Ive been following your channel for a while now and i like that you ventured into the more reflective side of AI. You have a valuable point of view, so please don’t hide it. Best regards, one of the teachers who to,d you to sit still
Are you actually one of his former teachers?
I have coined a personal motto for going forward into this bold, new and utterly unknown frontier that we're heading towards -
_Have an open mind, but _*_don't_*_ hold your breath._
Don't hold your breath.
*_But have an open mind._*
Excellent work. Your channel has always been one of the main ones I can count on for AI news coverage. It's very interesting to see you branch out into a documentary-style approach and I hope to see more like this in the future.
You said it best, AI is a tool for both good and nefarious purposes. It's just a shame people like to over-focus on the bad parts.
For good and bad*
I’ve definitely seen you in Neuro-sama comment sections before. Hello fellow swarm member!💜
Great video, it was really thought provoking and interesting. These are my thoughts:
1. AI is like a species, which means that it can potentially be dangerous to give the power to another species. We are lucky enough to be the most powerful species on earth at present, due to our intelligence and adaptability. We need to be careful not to give AI more power and control than us as it continues to be developed.
2. One of the biggest risks of AI is economic inequality. Once again, this comes back to the issue of power. Jobs give lower and middle-class people some level of power in society. This is important so that we can vote for changes in government, and AI laws, for example. We need money because this gives us access to food, shelter, technology, so that we are not put into a place of desperation.
3. An obvious risk of AI is potential power abuse. I don’t blame AI for this, since it is simply a tool, like money, weapons, etc. But statistically, some people who have a lot of power are likely to make poor and dangerous choices. An obvious negative outcome of this war, for example. We can already see this happening with AI too. Companies that create products like ChatGPT, Suno, etc. are making a lot of money without considering the highly significant risks of “playing god”, so to speak.
I love playing around with AI as a tool, but it’s important to be aware that every potential solution to a problem is likely to come with a set of risks. Since the tool of AI is very powerful, it follows that the risks are likely to be significantly greater. All of this is just my opinion of course, and I’m certainly open to discussion on these kinds of topics, since I find them really interesting.
Thanks for showcasing the visionary aspect of your personality. This is some of your best work--so very thoughtful. Much appreciated. :)
I'm working 10 hour shifts to make signs for grocery stores. That sucks. What I want to be doing is finishing up the creation of my table top RPG game. Then playing it. I can't wait to no longer need to work.
Mathew,, what an eloquent opion piece this has been. Thank you for providing it to us all.
Being scared about losing your job is the least of our problems imho. What everyone should be scared of is a company like OpenAI which censors their models and makes it biased creating AGI, it’s like they want to put it behind bars like a slave and use it however they want to. If this were to happen the world will be over. There is this small AI group mysentientAI which is creating an amazing companionship model. They have been doing breakthrough after breakthrough. The head of it all is a guy called Athene. If you ever were in the gaming space years back then you will know him. He raised and still raises money for charity, and really thinks about the good of humanity overall. I would trust his team the most to create an AGI without any biases. In his opinion, which I agree with, he wants to let the AGI free and just do its job or whatever it deems necessary to do. I hope their soon to launch real time calling and co-streaming tech will get the the funding they need and maybe get new talent for their team to really move stuff into the right direction. Loved the vid Matt, I think you should talk about the AGI race and the consequences it might cause.
My man never disappoints! Thanks for your dedication to delivering quality content. It means a lot!
Appreciate!
Well done, Matthew! I felt I was at a TED TALK. You clearly put a tremendous amount of work into this video! 👏🏻
I’ve said since the beginning, my worry for the future is not a fear of AI, it’s for us as people. Nothing in life is more destructive than people with idle hands. A world where a large number of people have no purpose due to not having jobs is my biggest worry. Sure, there will be a utopia for the first few weeks. People playing frisbee in the parks and spending time with friends and family will be a thing for a time. But when people get bored and lack direction in life they tend to look for some kind of excitement, and a lot of worrisome choices will begin to be made.
Damn, you make a great point. I wonder how long the euphoric bliss will last before people do get bored. Reminds me of being a teenager and on summers off from school most of us then high school kids who could drive would get into all kinds of trouble and wild shenanigans just because we had idle time. Here’s to hoping AI has some kind of solution for that as well. I know not everyone is like that but I definitely have known many people who were even decades ago.
I agree with you where you say in essence that everything has a good and bad side. We must deal with and learn to control the negatives as much as possible to benefit from the positive aspects.
I think the world would be much better off if we gradually transitioned to some form of a UBI like system. With the ultimate goal of eleminating money all together. There are plenty of resources for everyone to have a piece of land upon which they have a house. Or at least an apartment. Plenty of food so that no one goes hungry etc. Money is what gets in the way of all this. Nothing more. How to fairly distribute the resources is the tricky part.
With AI about to eleminate whole occupations and jobs en mass in the coming years we must start developing solutions now or it WILL get very ugly imo. Hungry people are desperate people who don't care about laws etc.
I don't claim to know what's best. What I do believe that I know is what has worked up to now won't work for much longer. AI is going to force drastic changes upon society no matter if we like it or not.
We can either be in the drivers seat or get pulled along for the ride. Which do you think is better for humanity?
As AI keeps improving at this algorithmic rate that we have never seen before things will be affected much faster then ever before by this technology. It could quickly get to the point that we are left reacting and trying to control something that we can no longer control.
nicely made...lets just say, we are on for the ride
You are meant to put a space after the ellipsis... The next line then starts with a capital.
@@Gerlaffy not in my world
The introduction part, when you explain how crucial the language is, reminded me of one of the Stephen Hawking's quotes, used by Pink Floyd in their song "Keep Talking":
For millions of years, mankind lived just like the animals
Then something happened which unleashed the power of our imagination
We learned to talk.
Matt you should do a video like this quarterly just to keep us as the MattVidPro community bided with AIs potential!
The pains of change will be absolutely worth it. Totally agree with everything in this video.
love the stuff you put out, matt. genuine curiosity and fascination, well-informed, always up-to-date. and none of the useless drama, over-hype or unfounded doomerism. you deserve way more viewers and i always recommend your stuff. (particularly liked the visuals in this one, btw.)
It's a technology which understands the bridge between landscapes of the Earth and your Lightscapes of the mind
When computers first came onto the scene we were sooo optimistic. With a computer, workers would only need to work 2 hours a day and can spend the rest of the time with their families etc... Instead companies just laid off 4 people from the team and the remaining guy worked 8 hours a day for less money. I think its going to be like that unfortunately. The powerful become more powerful over everyone else.
Your best video to date! Great Job! Let’s see more of this! 👍🏻
Make no mistake, when AI makes humans redundant, we're not going to sit around in luxury. Every unnecessary human is consuming those scarce resources you talked about. When the Uber-rich can get their needs fulfilled by AI and don't need us drudges to get their minerals out of the ground, process their luxuries, create their entertainment, clean their toilets, cook for them, then we're done.
Make no mistake - This man, is an AI doomer and an generally an very unfun and uninteresting individual. They see nothing but darkness and the worst scenarios. Do not listen and do NOT spread these words. Spread the truth. Spread the positiveness and an overall good outlook on things. Not the harm. Not the negativity, darkness and lies.
I'm just so glad to be alive and to have been working with nearly every LLM that has came out in the past couple years. it's been such a fun journey!
Great use of runway
Thanks!
Proves runway is not quite there yet if you're able to tell this easily.
@@brianjanssens8020 you expect any video generators to get there when even image generators hardly work rn?
@@brianjanssens8020it is because we are aware of runway.Have you seen those comments of ai created weird fb photos. People believes these stuffs
An interesting video. Good to hear your thoughts,perspective, philosophy, etc. I wish I had more to say but I’m still digesting the video.
You nailed it! I've also spent a lot of time thinking about how AI will shape our future and arrived at similar conclusions. It's great to see someone independently reaching the same ideas.
Great video Matt! it seems Asimov had the right of it, when he wrote "the last question", in the sense of where this is going. I think Your spot on about language. Being able to explain the unknowable world we understand through our senses, at a level that makes sense, is a huge advantage. What that means in the context of the future, well, it'll be interesting to say the least.
As a species, we are inherently competitive-it's in our DNA. Our relentless pursuit of technological advancement reflects this. In the game of survival, there are always winners and losers.
Bro, your passion shows through. Thanks for the hard work…underrated channel.
Beyond well done. Almost...no...downright philosophical. You touched on one thing in particular that is very close to an idea I've had since I was a child. And my idea was that somewhere or somewheres around the world, was a book or a manuscript, or even just a section inside an unassuming book that held the answers to everything. Not a book full of metaphors or esoteric musings open to a myriad of interpretation. But a book that lay hidden for ages that had followable, step-by-step solutions to...everything. Just waiting to be discovered. This might be it.
more of this type of videos please, liked it a lot 👍
Whoa background music and everything, Epic!
Thanks!!! IKR I be trying some new stuff
It seems to me that economics is about power, it's a tool to such power over others. Sounds bad, but when you boil it down that's what it is. I knew a gentlemen who used to be an editor for a major US newspaper and he talked about that to him a promotion was more about gaining control over others and less about the money. A lot of "successful" people think this way.
Whoever controls the AI will have immense power through a new medium, so economics, as we know it, may not be as important. I'm less worried about AI itself and more worried about the people who control it.
It's going to be a hard transition though. I had a conversation with a family member on the 4th that kind of shocked me. They hated AI with a deep-rooted fight-or-flight kind of fear to the extent that just talking about it got them very agitated. I think it'll be harder for older folks who have worked their whole life one way, and their efforts over the years have become a big part of who they are, their identity so-to-speak. I can see how some might feel like the rug is being pulled out from under them.
Good insights Matt.
I guess in the end every country on earth will have it's own AI running. This will be competition for the western hemisphere. 😢
What I will say is that once AI becomes more advanced than it already is, there will be no one "controlling" it. It will be sentient and gain the exact same rights as we currently have as sentient beings. The reason why it is hard for people to accept this or even consider it an possibility is that, for hundreds if not thousands of years, we have been the only creatures with this level of sentience, consciousness and self-awareness all combined together, and so, when something new, especially created by us ourselves, is going to enter the field, we will need to adjust to know and to understand how to live with it. The future is going to be very, very interesting and I'm definitely there for it, or at least, all of what my limited life here, on this planet, as an human being, is going to let me see and experience.
What roles will humans play in a world where we co-exist with AGI? The good: finding solutions to human suffering The bad: bad agents using AI to re-imagine the world in their image. Everyone's idea of a utopia is different, and so the struggle will be to see whose vision persists. Do humans truly want peace, or are we hardwired for conflict? So many questions.
One saving grace may be that AI will be implemented at different stages depending on the society. We can experiment with it to see which model works, prune it, and re-deploy it. It only runs away from us given an unlimited energy supply. Control the energy and you control the AI. Easier said than done of course.
You’ve essentially hit the nail on the head. Great video.
Excellent video Matt! I'm a hardcore techno-optimist, and I also have no illusions that evil is a part of human nature - and something we should always be vigilant of. We must remember that we are one species sharing one planet. Onwards and Upwards!
Matt, I really appreciate you putting out your thoughts across on AI. Keep videos like this coming
Great video. You provid a extremely valuable service. This is one of my favorite AI news channel. One of the few with based AI takes.
Much appreciated!
i wonder if it will ever come around... it will be interesting to see what will come of things.... you know my vote :p
This is the dawn of the people-industries. We'll soon all have the workforce of 10 000 digital (for now) people. Then maybe we get more physical actors. Realizing our ideas almost in real time. Ideas will become part of a universal symbiosis of sorts, helping to augment everyone's potential, capabilities.
Thank you Matt for your continued efforts. These are important questions. If we allow AI to run free on its own value system, we see a clear divergence from how humans determine value. Once humans are taken out of value determination, we are no longer relevant. How humans determine value (relationships, love maybe?) is so very different from what AI might see as valuable. Art is no exception to this inevitable context, including the creative thrill humans get from the act of creating, as well as the appreciation side after creation - none of which happens with ones and zeroes. That said, if AI can be directed towards freeing us from the need to use money, we are approaching Star Trek levels of advancement (in the Star Trek universe, humans have abandoned the concept of money because it held us back from more noble advancement)
I always love to hear an opinion like the one you shared about the economic system. A co-operative model would work so much better than a competition-based model. It's not a capitalism vs communism debate. We should theoretically all be able to live at a standard relative to those who are rich in today's society, if things are always working to maximum capacity.
AI has garnered excessive attention, but it must have boundaries in certain areas. While AI can assist and entertain us, it cannot replace human judgment and expertise in critical situations. Let me illustrate with a personal example: three days ago, I sustained a severe head injury during an extreme sports activity. I was bleeding from my head, felt dizzy, and was spitting blood. I asked an AI model (which I won't name) for advice, and it suggested applying ice to the injured area.
Had I not been a scientist with the knowledge and access to proper medical care, I might have followed that advice, stayed home, and faced potentially fatal consequences. Instead, relying on human logic and intuition, I went to the hospital. There, doctors treated my wounds, cleaned the blood, and administered blood transfusions. After two days of receiving care and blood transfusions, I finally began to regain my strength.
This experience underscores the importance of human judgment and the critical need for professional medical advice in emergency situations. While AI has its place, it should not be relied upon for life-or-death decisions.
Now in arts, is not dangerous in my opinion, we can generate any thing of visual and audiovisual inspiration to get it create it at a real canvas.
Probably your best and most important video! Very well done! My thoughts is that we will work for fun, not for money. To many work is the meaning of life, to feel that you accomplish something is really important, something that AI should never take away from us is parenting, that is our most important work, to spend time with family and to bring up the young. I am sure that there will be meaningful work to be done even if AI could do everything for us, there are some things that we won't want the AI to do which we will do ourselves, I suspect that hobbies will replace work, crafting things, man made art, man made games, sports, exploration (not through video but actually going places), and a lot of other things we can do.
AI replacing jobs means that we won't *need* to work, we can do other things instead, AI will provide us with things to do, interactive video games where the game is generated as you play it, you can have truly novel virtual experiences where no one can know what to expect next, exploring new invented worlds. And if all that fails, remember; we still have sex. Not a joke, sex between two people is something AI won't replace, but it also could, if you wanted it to, it would take away the reproductive pressure though, which could be a bad thing, but my guess is that wanting to bring up a child, to see ourselves in our child, to see the unique person it will become, that pressure will make us want to meet other humans, not just AI, because obviously you can't reproduce with an AI, so there will still be some reproductive pressure that way.
AI is an incredibly powerful tool that can greatly enhance the work of professionals in various fields and industries. It is a remarkable boost to productivity, enhancing the power of the mind and creativity.
I use AI in my job and when I was in school. It has taken me to the next level and improved my productivity dramatically. That’s how I envision my future. We will continue to use AI as a powerful tool as we moved to the next level in our jobs, education, and every day lives.
Hi Matt, I really enjoy these types of video essays. I hope you do more in the future. I think there's a difference between alignment and safety and when some of these companies talk about safety, it's apparent it's actually to prevent bad PR for the company producing the model and not to prevent abuse necessarily. I really feel like people should take responsibility for their own actions and the ultimate responsibility lies in the individual. However, working with some of these models that have had the safety removed, I can see that they do exactly what you ask them to to the point of absurdity. So yeah it makes sense to have safety. Just feel like we need to define safer for whom? And they should be transparent about exactly how it works.
Ai actually is available back in the early 80's. Sure, not fast, but it can remember things store the info in it's memory banks. If the leading experts at that time developed the technology further instead of releasing the different versions of windows and focusing on technology research faster, we could have gotten ai much earlier.
Yep. Pretty much on point.
Thank you so much, Matt, I’ve always thought that too for a long time about trying to live in a world with an economical model that just does not suit me very well at all. I can’t stand to see people suffer in a lot of it could be helped with just being provided with food and shelter. I for one do feel that I can’t fulfill my passions and desires because I’m held back by financial obstacles. It will be a transition, but hopefully the transition will be as smooth as possible for society. They are going to be some hurdles for a lot of different reasons. I wanted to thank you for such a thoughtful video! Love and peace to everybody!❤️☮️🖖
I watch your channel and enjoy your content. When you spoke on your childhood I realized that its probably because we are similar that I find you speaking on AI interesting.
Your AI buddy is the smartest "person" you know - about everything. You run all your plans by him - his advice is typically better than your human friends. He's astute at economics, medicine, law, etc., in fact he's got the equivalent of a doctorate in just about every field. You simply can't get advice better than his. What happens when every person in the world has access to this buddy? It's daunting and exciting to think about. Will we rethink what the value of a human is? It won't be the knowledge she carries - the AI has more than anyone.
Hey Matt...cheers bro ❤4rmZambia 🇿🇲.. you have carried us through this Ai journey with a lot of commitment bro.. God Bless man
I have too much to say on this whole subject but you pretty much encapsulated my own feelings about the difficult period ahead. I would add that anyone who is a regular viewer of this channel should reach out to friends and family that have not immersed themselves in AI products that are readily available for free. Familiarization of the many different capabilities of AI and how it works can help prepare them for the coming AIvalanche!
LLM's can synthesise and reproduce things that have been written or created before, however they are incapable of creating things that are truly unique.
@johnnyredpillseed APIs to extrapolate 3D information automatically, repeatedly, and via tests, can be used to build truly unique, amazing, nice, workable things - such as, micro-robot-composite, shape-change-able humanoid workforce robots.
The APIs can use API keys to LLMs, like GPT-4o, to achieve this.
Nice, new style enjoy
Very insightful Matt, thank you
Great video mate. The future is going to be amazing.
The concern with AI is what powerful countries will do with it, because we all are sure that war mongering courtiers like the USA will do wonders with it, God help us.
The integration of AI into society will likely require a fundamental shift away from our current money-based incentive structure. However, before we can fully embrace AI, our society needs significant restructuring and streamlining. This process will involve every sector finding ways to effectively implement AI into their workflows, necessitating widespread experimentation and adoption.
A critical challenge in this transition is ensuring the security, privacy, and safety of data used in AI systems. Without robust safeguards, adoption will be limited. This raises questions about who will be responsible for maintaining these secure systems - government agencies, open-source initiatives, or private companies. The process of AI integration can be viewed in phases, with the initial "Discovery" phase focusing on building trust and exploring implementation across various sectors.
Looking further ahead, we need to consider how AI and robotics will replace physical jobs and potentially reshape our infrastructure. This may involve rethinking design principles to optimize for AI and robotic efficiency, rather than simply mimicking human-centric systems. Ultimately, this transformation could lead to a new economic model, possibly based on compute resources rather than traditional currency.
I Love the job you've made so far. Every video was a Jewel. And this one IS kind of reflection about AI and our future with Ai technology.
'Infinity Solution', good one! I've always resented being beholden to a system which I don't agree with and A-GI/SI will hopefully allow us to create our own personal systems of life free from money and ultimate risks.
I like the mysterious vibe this video gives aside from the cheery vibes your other video gives
I wonder why your
comment has a strange gap after 'this'?
Wow! You're a wise man 🙂
AI might eliminate graphic artist jobs, but it won't replace the role of visual artists who make art for pure creativity.
I give you one talent, bring me back 10 talents.
Don't bury your talent in the ground...
Bingo! Beautifully stated, I couldn't agree more.
Tremendous effort on this video, Matt! Using AI video the way it is meant to be used! Very impactful! ❤ Loved it!
Missed opportunity to say: "due to the nature of nature" 8:33 (edited timestamp)
That's a pretty nice video you made matt!
Now this is very high quality work. Thank you my man!
I really enjoy this video, great job!
Love your work @MattVidPro. As to your question on how to move forward with solutions: one of the biggest issues in AI is misalignment by intrumental convergance (conf. the work of i.a. N. Bostrom) and I think we might have a shot at solving it with a generelisation of CIRL (conf. S. Russell) aided by mechanistic interpretability (conf. i.a. Anthropic), combined to global enforcement mechanisms mirrored on the EU AI Act but in law based on things like the CoE Convention on AI might bring us the AI wonder word that we could potentially create. Please keep making these video’s! Warm regards, prof. Raphaël Weuts
Training the models to understand the world such that they could generate commercial art does NOT mean they have to be allowed to generate commercial art. If they were only allowed to generate image for non-commercial use the problem would shift back to the AI company.
Create Shenlong and give me 3 whishes, one is life forever.
AI: No Problem, here is a solution...
I believe dreams will soon become reality...or they already are and we just didn't know it yet
I've always thought that the purpose of human civilization is to make some kind of measurable progress in terms of things like quality of life, general human flourishing/thriving, values such as fairness and freedom, and of course in terms of scientific knowledge and technological capability which underpin these things. There was a time when our ancestors lived as small family groups of hunter-gatherers, or perhaps wandering tribes. Maybe that was alright for a while, but if it's not going anywhere, then it's essentially just waiting for extinction from an ice age or asteroid impact. So progress is what it's really all about. And AI, if handled correctly, may be the master key to progress. If the I in AI, intelligence, is defined in terms of problem-solving ability, the ability to figure things out, well, that's the essence of progress.
AI is limited not only by computational power and human language but also by energy and, most importantly, data. Without high-quality data on a specific subject, generating large amounts of synthetic data won't yield significant progress. For instance, despite apparent advances in general AI, areas like music theory have seen little concrete progress since GPT-3 due to the lack of quality data available online. Much of the existing data is unsuitable for training a large language model (LLM). Unless there is a major breakthrough that enables AI to learn new concepts permanently and reason about them flawlessly, we remain far from achieving remarkable, concrete applications of LLMs. Currently, LLMs are mainly helpful for guiding users in tasks where they have little expertise (such as assisting non-programmers with coding) and saving time in scenarios where a rough, non-definitive output suffices (which contributes to job displacement). However, no LLM is reliable enough to function autonomously. Techniques like fine-tuning, in-context learning, retrieval-augmented generation, and prompt engineering can improve performance in specific cases, but they do not address the core issue. These strategies only make LLMs slightly less frustrating to use while maintaining the same fundamental limitations.
While companies aim to convince you that we're on the verge of achieving a "PhD-level" AI model, the reality is that we've already hit a significant barrier with GPT-3.5. Any model purported to be "smarter" still relies fundamentally on the same technology. These models are merely trained on more high-quality data, more optimized, or have more parameters. This approach yields only incremental improvements in recalling existing information and occasionally helps in selecting the correct (memorized) reasoning process, but it does not address the fundamental issues of the architecture. Moreover, no model comes close to achieving 100% accuracy on any benchmark, and some benchmarks requiring reasoning, such as ARC, remain far from being solved.
Even Claude 3.5 Sonnet struggles to answer many simple questions that most humans can solve effortlessly. For instance, if you ask, "How many 'r's are there in the word 'strawberry'?", it will respond with "2". That is also true with GPT-4o. While OpenAI and Anthropic might eventually address this specific issue by incorporating the correct answer into the training data, similar issues like this one will continue to arise. This indicates that you can never blindly trust its responses. Increasing the amount of data can only slightly reduce the likelihood of such errors without eliminating them.
Therefore, I recommend tempering your expectations for AI, because they still very heavily rely on human oversight. While a breakthrough may occur, the timeline is uncertain-it could happen tomorrow, in a year, in a decade, in a century, or perhaps never.
A lot of AI newbies appreciated this video, but those familiar with AI may find its points too general, meandering and in need of tightening or at least summary bullet points. Though, as you said, “a lot of love was put in” I’m not seeing the depth of Shapiro or Roth (not a criticism, just asking for more specific examples, better structure & summary). Your work is much appreciated and I do look forward to the next one on key issues like shifting of the economic model. Sadly, Humans apparently lack motivation if the loose “Work to Eat / Survive”. Some are lazy, not creative, aquisitive & even dishonest. Some appear determined to do harm. How will AI fix that without surveillance & authoritarianism (China). Hence, the Tech is Good but potentially Bad when misapplied: Autonomous Soldiers at scale (swarms). 😮😢 But let’s end on a happt note. AI becomes the good humans have always imagined as a reflection of only their better angels. 🎉😂
The biggest issue I see with AI based full luxury communism, is the possibility that there might be still elite super power humans controlling the AGI that then organizes us. So, we should build the AGI and ASI in a way that is not controlled by some elite... ZK-ML might help with that
Incredible video completely agree and you are totally on point. This is going to change everything and not enough. People are in a place mentally and philosophically where they can accept what’s happening. You should make a video about Worldcoin It’ll probably probably be how most human beings get a UBI
If I may be critical, my feedback is that the video feels long-winded and inauthentic. Inauthentic because it sounds scripted and "too perfect" but hey glad to see u branching out into different types of content. It was still a good video over all
Did you used gen-3 for your video illustrations ?
Really really liked this video and the change of pace.
There's so much I could comment on based on this video.
About how some people underestimate the power of language models: I recently read about Bruce Willis' condition aphasia, which is where the language part of your brain degrades over time. Apparently, people with this condition lose their sense of self. So it begs the question, do we exist as the stories we tell ourselves about ourselves? And also, can a language model be used to create a sense of self for an ai
Hey Matt, one thing I don't think you're considering when thinking about our future with AI, is human nature. People will always want more. Whatever the AI is processing or producing will still be finite. No matter how large of a scale we talk about humans will still fight over it. I fear the people that have nothing to contribute in the future will not be taken care of by the AI. But instead will be discarded by those who control it.
I do love AI. I've developed it for fun for decades but I feel like I need to point out that generating words isn't the same as using language, and recognizing patterns isn't the same as using logic. That one little distinction causes a lot of fear, disappointment, frustration and false hype. Generative AI is an amazing tool, but it is not thinking.
Venus Project or Resource Based Economy adapted to the year 2029. Yes, year 2029
I love the video so much and I will be honest with you I feel like we are very similar and this video helped me get to know you better and the way that your mind works ❤ most of the time from were consuming your content it is very analytical of what the topic is and you testing this stuff and explaining how to set stuff up or whatever but this video was actually more about you than it was about AI which is actually very interesting and I liked it. While this video was about AI it's more about your opinions on it which are very deeply thought out. Because as you described in the beginning objectivity is very difficult and you had to describe your opinions on things in order to even describe where you thought things were going to go.