Always a great range of issues and articles covered here with genuine grace and good will, and novel but appropriate responses for us to react to and think on without feeling steered in any specific direction... Fast becoming a favourite channel o' mine. Keep up the great work.
Thank you so much. Not having an agenda or strong opinions makes growing this channel hard but the fun part for me is the curiosity and pondering so that's mostly what I do. If I can build an audience of people that appreciate that I'm gonna be beyond happy! Thx for the support and insightful comment.
Oh my gosh! Another super chat! Man it feel so good to look down and see these! Thank you. Let me know if you want me to cover any specific topics or anything.
A robotic snake dress? This has got to be the craziest timeline available. 🤔 🤡🌎 Dylan I cannot thank you enough for your show. You have condensed my AI content down to just you and Matt Wolf once a week. I used to watch five or six different channels every week and now it's pointless. Thank you so much for your hard work and effort. Thank you so much for your perspective and feedback. I absolutely love this channel and I'm doing everything I can to promote you and share this amazing entertaining and educational content. Just clipped a few parts and made a new short. 👍🏻💓
@@dylan_curious Not that I need any help with talking to strangers, but I think just about any of your videos could give me a full night of crazy futuristic ideas to talk about. Love your work brother! 😀
Well, you just confirmed what I've been thinking for the last months. I shouldn't care about my own judgement on reading books about things I don't know entirely. Active inference texts, here I go! This episode was really cool, as usual. Thanks for the amazing work as usual, mate! Cheers
Your comment reminded me I need to talk about active inference and the "Free Energy Principle". I better start listening to some people talk over my head about it now.
@@dylan_curious both are very important topics and very relevant so we should get to it. The machine learning street podcast channel has it covered, I really recommend you to watch their videos
The patches of pixels - recalling a few years ago more simple machine learning would use the outline of the main image so could tell yoga poses for example , or in my data science project, classify a pose for a dog in their adoption pic to see if the pic influenced adoption more than the words (btw at the time during covid data showed it was the words but of course data was skewed due to shortage of dogs cuz everyone wanted companions )
when you train an image/video generation model, the visual data is tagged... with WORDS xD instead of learning to equate patterns between shorter prompt input like "song lyrics about cats" and the complex patterns of songs, what are lyrics, and what are traits of cats to construct a new complex text string, it's correlating input to text+pixel patterns. over time it can correlate the simple input -> simple tags -> complex patterns of gradients, colors, and complex geometries
So let me know if I got this right. Does word "cat" as a word token sit in a space around other pixel patern tokens in a position that places it some distance from many linear combinations of learned patterns?
@@dylan_curious mostly, yes, but with modern attention and latent diffusion... not sure it counts as a purely linear feature space anymore. original alexnet, and earlier text to image based entirely on convolutional networks, yes.
A little late on the comment, but excellent video. man. The study about individual neurons having active influence is astounding! I'm a firm believer that free will is an illusion, and this brings much deeper scope to the subject. I wonder what Dr. Sapolsky's thoughts on the research will be.
I lean towards free will being an illusion also, but I’m not 100% on it. There could be some kind of emergent control when system has become complex enough where everything feels like it’s bottom up, but there is a top down intelligence that is more than the sum of its parts. But I would not bet on it, probably everything‘s just playing out the way it plays out.
@@dylan_curious Emergence is a concept I love exploring, especially in regards to consciousness. I just see emergent consciousness as a system of complex observation of our mind's virtual model of the surrounding world, which should leave no room for free will. However, I'm hoping to be proven wrong! That would be nice! If not, I'm betting attainment (or perhaps simulation) of "true free will" to be one of the ultimate goals of mankind. Thx for responding, man!
30:46 Scientists are like the guys the discover the glitches speedrunners use; from the LHC smashing things so fast that they briefly glitch revealing underlying hidden rules, to astronomers studying glitchy regions like the surroundings of blackholes, magnetars etc, trying to figure out the hidden ways whatever darkmatter is influences the behavior of galaxies, peering at the edges of the draw distance by looking deep into the beginning of the Universe; and so on. Maybe with ASI we might be able to TAS it...
Ill keep preaching- brains arent sublime. Using tissue and stem cells as a dual interface will develop a whole new field of research. We gotta move away from scare trains around scifi and essentialist perogatives; instead of worrying about if this robot gains 'conciousness' we should be worrying about how biases are programmed into the machine. Whats dangerous isnt necessarily the output or existential status- its the data we give it. Scares shouldnt really be about consciousness- something like this is infinitely more ethical than something like monkey research.
I have heard some theories that as you become more intelligent, you naturally become more empathetic also because you understand all the things around you, but I totally agree with how we should be worried about today is what input data warrior giving it what the goals are. Great comment.
@dylan_curious Empathy vs sympathy is a huge thing too. AI models are inherently incapable of empathy; sympathy then should be seen for what it is, and adjustments made accordingly. I might dispute that intelligence equates to more empathy, if we are talking sheer compute the potential for sympathy increases, but once again it's entirely based on the data given. Some of our most intelligent leaders are psychotic, look at the field of capitalist CEOs, the over representation of sociopaths in leadership, look at Musk. You don't raise the kid right, you don't give the right data, and it doesn't matter how smart or stupid they are. Goal oriented existence, how we accomplish them is all based on what we've learned.
Eckhart Tolle says we need difficulties in life. If things are too easy it's not fullfilling. Even game makers know this: when you make a game you want it to be not too easy, because people will get bored and not too hard because people will give up on it, because it feels like they can't win. So yes, I worry about that too, Dylan.
You know I read a couple of his books years and years ago and I remember thinking about them a lot when I first read it. You just rekindled a lost memory. I should reread those books.
@@dylan_curious Oh that's great! I had a video of him talking about this pop up in my video feed, out of the blue, today. I hadn't watched his videos in a while. I like the way he tells it, I should probably take the time to read his books too, I've only watched his videos so far.
I might have missed it but did Dylan do his review of how close we are to AGI, from, Dr Alan Thompson? Cuz we were at 75% for very long time but I think Dr Allen Thompson raise it up to 82%! So that's 18% left to achieve AGI. If it's an exponent increasement then we just one step left?
I check it every time I film, but I only bring it up when there is an update and it's still at 75% I feel like it will still be a little while because he's big on embodiment and we need to wait for robotics to catch up but once robotics gets there then it's all gonna happen at a snap of a finger.
30:50 Are we in a simulation Buddhist, Yogis, Krishnas, Non dualist,Sumerians: Yes we've been trying to tell you this and we even provided instructions on how to access the settings. lol
Dylan! You should do a segment on this: ruclips.net/video/gbTBoPdcsKw/видео.html I don't know why no AI channels are talking about it. It's the most amazing thing: You create an artwork in your mind, and the AI can see it directly in your brain and make it real! And by the way, the guy whose RUclips channel it is (his name is Cody Rall) seems to be practically the only person on RUclips with a channel about neurofeedback, which is super-interesting.
"I'm worried the world will have too much pleasure in it." Solution: learn more history. You'll very quickly realize that this is the best problem to have in all of history. By a lot. We've had many cultures with leisure classes and they were *very* fine with it. :D
LOL. Could you imagine if I say "yes Grimlock, it's extinction time" then go into some dystopian concept. If I remember I might just do this to be funny. Thx for the idea.
The fact that I keep hearing people leave Twitter. I wouldn't doubt that dead internet hypothesis tbh. And definitely wouldn't put it past Musk to artificially boost interaction to keep the few people left. 26:46 I always liked that generative agents look like Pokemon trainer sprites lol. 😂
I never noticed that before, but you're right it does look like Pokémon trainer sprites! Yeah, and we never know what are bots and what's human now that LLMs are so convincing.
Do you think you could try and make the title transfer cards less sexualize? Its not that bad, but it does bring some discomfort if youre wanting to hold the most mainstream appeal. Maybe just a personal gripe- the last few videos ive noticed theyve been a bit more sexualized.
The article was published in the "South China Morning Post" which I believe is run by the state. But it seems like a real thing. Google "Chinese scientists create robot with brain made from human stem cells" and you can see the autor Victoria Bela. Could be fake not sure. I thought it was real.
@@dylan_curious I'd like to believe it. But the tissue above the liquid would dry out very fast, wouldn't it? It's like the story of the first super conductor at room temperature or of the first cloned human, which were scams. This looks like another one to me. Hopefully Sabine Hossenfelder will make a video about it.
@@RodsFromGod_26 Ok, ok. ok. dang. now I see it your probably right. I don't know what brain grown from stem cells looks like? I mean it would not have wrinkles because it's not hitting the top of the skull.
You are definitely smarter about this stuff than I am - yet you don't 'explain it down' to me - and your unpretentious for sure. Hey and if I don't quite get it - that's okay.
Deep down I feel like I am always in over my head so don't worry about it. I still ask ChatGPT all the time to explain basic stuff I forgot and want to see if I know correctly and it's embarrassing how much I forget.
Thanks!
You are so kind! I appreciate the support so much!
Great work.
Always a great range of issues and articles covered here with genuine grace and good will, and novel but appropriate responses for us to react to and think on without feeling steered in any specific direction... Fast becoming a favourite channel o' mine. Keep up the great work.
Thank you so much. Not having an agenda or strong opinions makes growing this channel hard but the fun part for me is the curiosity and pondering so that's mostly what I do. If I can build an audience of people that appreciate that I'm gonna be beyond happy! Thx for the support and insightful comment.
Another dose of MAD news. Cheers, man and hello from Taiwan! 🇹🇼🇹🇼🇹🇼
It’s a MAD world. So yeah… MAD news.
Love watching your videos Dylan!
I appreciate having you here! Thx man.
Same here ❤😊
Thanks you!
Oh my gosh! Another super chat! Man it feel so good to look down and see these! Thank you. Let me know if you want me to cover any specific topics or anything.
Love the vid man.
Always a please to watch them.
love your jokes. good video
Glad you like them! I genuinely have fun filming these videos so hopefully it comes off that way.
Agreed! Going to the gym 😂
good topics and discussions
A robotic snake dress? This has got to be the craziest timeline available. 🤔
🤡🌎 Dylan I cannot thank you enough for your show. You have condensed my AI content down to just you and Matt Wolf once a week. I used to watch five or six different channels every week and now it's pointless. Thank you so much for your hard work and effort. Thank you so much for your perspective and feedback. I absolutely love this channel and I'm doing everything I can to promote you and share this amazing entertaining and educational content. Just clipped a few parts and made a new short. 👍🏻💓
But you know that dress is a great talking point at parties!
@@dylan_curious Not that I need any help with talking to strangers, but I think just about any of your videos could give me a full night of crazy futuristic ideas to talk about. Love your work brother! 😀
well done
Awesome
Thanks for taking time to leave a comment.
Woww, this was great. Thank you for your great videos. You spoke very well about my article. , thanks Dylan.
It's so awesome to have an author in the comment section! I really appreciate the effort you put into making your content. Happy to build on it.
@@dylan_curious Thanks 🙏
Well, you just confirmed what I've been thinking for the last months. I shouldn't care about my own judgement on reading books about things I don't know entirely. Active inference texts, here I go!
This episode was really cool, as usual. Thanks for the amazing work as usual, mate! Cheers
Your comment reminded me I need to talk about active inference and the "Free Energy Principle". I better start listening to some people talk over my head about it now.
@@dylan_curious both are very important topics and very relevant so we should get to it. The machine learning street podcast channel has it covered, I really recommend you to watch their videos
00:50 - I remember Krang. I wish you'd have referenced Motoko from Ghost In The Shell. However, Krang makes sense.
I don't know Motoko. I will need to look into that character. Is Ghost In The Shell worth watching?
The patches of pixels - recalling a few years ago more simple machine learning would use the outline of the main image so could tell yoga poses for example , or in my data science project, classify a pose for a dog in their adoption pic to see if the pic influenced adoption more than the words (btw at the time during covid data showed it was the words but of course data was skewed due to shortage of dogs cuz everyone wanted companions )
when you train an image/video generation model, the visual data is tagged... with WORDS xD instead of learning to equate patterns between shorter prompt input like "song lyrics about cats" and the complex patterns of songs, what are lyrics, and what are traits of cats to construct a new complex text string, it's correlating input to text+pixel patterns. over time it can correlate the simple input -> simple tags -> complex patterns of gradients, colors, and complex geometries
So let me know if I got this right. Does word "cat" as a word token sit in a space around other pixel patern tokens in a position that places it some distance from many linear combinations of learned patterns?
@@dylan_curious mostly, yes, but with modern attention and latent diffusion... not sure it counts as a purely linear feature space anymore. original alexnet, and earlier text to image based entirely on convolutional networks, yes.
Was there as much Ken burns effect in your other videos? I noticed it heaps in this one.
Depends on how much time we put into editing. Which one did you like better?
Without the sour the sweet is just not as sweet. Vanilla Sky. 👌
A little late on the comment, but excellent video. man. The study about individual neurons having active influence is astounding! I'm a firm believer that free will is an illusion, and this brings much deeper scope to the subject. I wonder what Dr. Sapolsky's thoughts on the research will be.
I lean towards free will being an illusion also, but I’m not 100% on it. There could be some kind of emergent control when system has become complex enough where everything feels like it’s bottom up, but there is a top down intelligence that is more than the sum of its parts. But I would not bet on it, probably everything‘s just playing out the way it plays out.
@@dylan_curious Emergence is a concept I love exploring, especially in regards to consciousness. I just see emergent consciousness as a system of complex observation of our mind's virtual model of the surrounding world, which should leave no room for free will. However, I'm hoping to be proven wrong! That would be nice! If not, I'm betting attainment (or perhaps simulation) of "true free will" to be one of the ultimate goals of mankind.
Thx for responding, man!
30:46 Scientists are like the guys the discover the glitches speedrunners use; from the LHC smashing things so fast that they briefly glitch revealing underlying hidden rules, to astronomers studying glitchy regions like the surroundings of blackholes, magnetars etc, trying to figure out the hidden ways whatever darkmatter is influences the behavior of galaxies, peering at the edges of the draw distance by looking deep into the beginning of the Universe; and so on. Maybe with ASI we might be able to TAS it...
That's a fun way to look at it and you're totally right!
Dylan - you’re looking straight fire atm 💇♂️🔥
I just don't want a real-life Adam Smasher to exist. That would be bad for everyone.
Ill keep preaching- brains arent sublime. Using tissue and stem cells as a dual interface will develop a whole new field of research. We gotta move away from scare trains around scifi and essentialist perogatives; instead of worrying about if this robot gains 'conciousness' we should be worrying about how biases are programmed into the machine. Whats dangerous isnt necessarily the output or existential status- its the data we give it. Scares shouldnt really be about consciousness- something like this is infinitely more ethical than something like monkey research.
I have heard some theories that as you become more intelligent, you naturally become more empathetic also because you understand all the things around you, but I totally agree with how we should be worried about today is what input data warrior giving it what the goals are. Great comment.
@dylan_curious Empathy vs sympathy is a huge thing too. AI models are inherently incapable of empathy; sympathy then should be seen for what it is, and adjustments made accordingly. I might dispute that intelligence equates to more empathy, if we are talking sheer compute the potential for sympathy increases, but once again it's entirely based on the data given. Some of our most intelligent leaders are psychotic, look at the field of capitalist CEOs, the over representation of sociopaths in leadership, look at Musk. You don't raise the kid right, you don't give the right data, and it doesn't matter how smart or stupid they are. Goal oriented existence, how we accomplish them is all based on what we've learned.
Eckhart Tolle says we need difficulties in life. If things are too easy it's not fullfilling. Even game makers know this: when you make a game you want it to be not too easy, because people will get bored and not too hard because people will give up on it, because it feels like they can't win. So yes, I worry about that too, Dylan.
Simon says move two steps back.
@@ericpatterson8432 How do you mean?
You know I read a couple of his books years and years ago and I remember thinking about them a lot when I first read it. You just rekindled a lost memory. I should reread those books.
@@dylan_curious Oh that's great! I had a video of him talking about this pop up in my video feed, out of the blue, today. I hadn't watched his videos in a while. I like the way he tells it, I should probably take the time to read his books too, I've only watched his videos so far.
I miss the red eye. Can you AI one into the videos?
Only with AI. I hope to never pop another blood vessel in my eye. Knock on wood.
I might have missed it but did Dylan do his review of how close we are to AGI, from, Dr Alan Thompson? Cuz we were at 75% for very long time but I think Dr Allen Thompson raise it up to 82%! So that's 18% left to achieve AGI. If it's an exponent increasement then we just one step left?
I check it every time I film, but I only bring it up when there is an update and it's still at 75% I feel like it will still be a little while because he's big on embodiment and we need to wait for robotics to catch up but once robotics gets there then it's all gonna happen at a snap of a finger.
06:01 Jen be 😨
So true! Plus Grimlock was way to expensive to not catch!
30:50 Are we in a simulation
Buddhist, Yogis, Krishnas, Non dualist,Sumerians: Yes we've been trying to tell you this and we even provided instructions on how to access the settings. lol
Honestly, the idea has always resonated with me.
Dylan! You should do a segment on this: ruclips.net/video/gbTBoPdcsKw/видео.html I don't know why no AI channels are talking about it. It's the most amazing thing: You create an artwork in your mind, and the AI can see it directly in your brain and make it real! And by the way, the guy whose RUclips channel it is (his name is Cody Rall) seems to be practically the only person on RUclips with a channel about neurofeedback, which is super-interesting.
KRang!
What a throwback.
❤❤❤❤❤
Thanks for always leaving a comment!
Shaped by your unconscious not your motives. Your motives are the consciousness explaining its subconscious actions after the fact. #nofreewill
This is something I 100% agree with. We confabulate our motives and split brain studies is the evidence of this.
"I'm worried the world will have too much pleasure in it." Solution: learn more history. You'll very quickly realize that this is the best problem to have in all of history. By a lot. We've had many cultures with leisure classes and they were *very* fine with it. :D
I can only imagine how many people in history would want to have that problem. Even animals. Thanks for the reminder.
@@dylan_curious Thanks for the song, btw :)
❤
Back at ya
Instead of correcting grim lock each time for the news I would save the dystopian stories for extinction time.
LOL. Could you imagine if I say "yes Grimlock, it's extinction time" then go into some dystopian concept. If I remember I might just do this to be funny. Thx for the idea.
06:24 OCP a subsidiary of CCP?
The fact that I keep hearing people leave Twitter. I wouldn't doubt that dead internet hypothesis tbh. And definitely wouldn't put it past Musk to artificially boost interaction to keep the few people left.
26:46 I always liked that generative agents look like Pokemon trainer sprites lol. 😂
I never noticed that before, but you're right it does look like Pokémon trainer sprites! Yeah, and we never know what are bots and what's human now that LLMs are so convincing.
Do you think you could try and make the title transfer cards less sexualize? Its not that bad, but it does bring some discomfort if youre wanting to hold the most mainstream appeal. Maybe just a personal gripe- the last few videos ive noticed theyve been a bit more sexualized.
Yeah, I definitely want bring AI to a larger audience so I need more mainstream appeal. I appreciate the feedback
If you aren't working your azz off then you are VERY privileged.
Why is nobody commenting the obvious?? This brain robot is totally fake. WTF?
The article was published in the "South China Morning Post" which I believe is run by the state. But it seems like a real thing. Google "Chinese scientists create robot with brain made from human stem cells" and you can see the autor Victoria Bela. Could be fake not sure. I thought it was real.
@@dylan_curious I'd like to believe it. But the tissue above the liquid would dry out very fast, wouldn't it? It's like the story of the first super conductor at room temperature or of the first cloned human, which were scams. This looks like another one to me. Hopefully Sabine Hossenfelder will make a video about it.
IKR? The "brain" looked like a blob of Play Doh.
@@RodsFromGod_26 Ok, ok. ok. dang. now I see it your probably right. I don't know what brain grown from stem cells looks like? I mean it would not have wrinkles because it's not hitting the top of the skull.
You are definitely smarter about this stuff than I am - yet you don't 'explain it down' to me - and your unpretentious for sure. Hey and if I don't quite get it - that's okay.
Deep down I feel like I am always in over my head so don't worry about it. I still ask ChatGPT all the time to explain basic stuff I forgot and want to see if I know correctly and it's embarrassing how much I forget.