Yo!! Thank you for the super thanks! You are getting pinned! And thanks for the feedback about length I have experimented with some other video length and I think for now I'm gonna stick with the 30min thing. It's what Matt Wolfe is doing and he is the top dog in this space but if if I get a plateau and need to do some changes, I will consider. You rock for leaving a comment and supporting the channel.
@@dylan_curious Yeah you've got a good format so that if people wanna just check it out for a bit to catch the news, they can, and if they wanna keep watching then that works too. Still, 5mins chunks of the individual stories might be a good way to keep the 5miniters happy and generate more content for basically no extra work
I absolutely loved this episode. Intro was perfect, reading people's replies to your topics, your thoughts and wisdom, interesting and relevant AI and tech news. You are doing an amazing job, just continue doing what you're doing and make sure to enjoy it. The future looks bright for Dylan Curious! 👁️🔮😎🌟
Contraversial opinion about AI I shouldn't say out loud: policing information is not safety. Breaking into a bank is illegal; knowing how to is not. We shouldn't be so cavalier about embracing thought control in the name of "safety"
I get your point. The tricky part is sometimes information is synonymous with capabilities. Like the lock maker does not want everyone to have the knowledge to pick a their locks. But I guess you're saying would the world be better if everybody did have that knowledge and then anybody who wanted to invent a secure lock had to come up with a new design even they themselves couldn't break into?
@dylan_curious Partly, by having transparency, we can build better defences. Partly it's a matter of the lesser evil: if OpenAI says they are building cognition for everyone, but unilaterally declare that certain thoughts are not allowed, then that presents a huge risk in itself that may be greater than original problem. Having said that, there are abuses that need to be prevented, such as deepfakes. But I think we should emphasize solving the problem of inappropriate *action* rather than inappropriate *thought*. But there is nuance.
Or here is a concrete example: most art museums contain art that would be blocked by most generative AI companies. I have a photographer friend who tries to do interesting art in Midjourney but can't even for very tame subject matter. So what does "AGI for everybody" mean when certain topics are blocked unilaterally. There probably are things that should be blocked, but who decides? Right now we just accept that things are blocked without asking if that's okay or who should decide.
There's gonna reach a point where anyone would be able to just go "Hey ChatGPT, print me a batch of hyper-COVID in my outdoor gene printer you built last week and hook it to the sprinkler". How do you feel about that scenario? Should downloadable-doom (not the game) be something every individual on the planet can have access to, do you trust every single individual on the planet with that power?
I really feel like you are hitting your stride with the channel Dylan. There are dozens of channels that jump on the latest news. Some of them even stream the events live and have commentary on the fly. You have a special knack for finding nuggets of weirdness. Like paging through a Guiness Book, or something. Love the humor as well. I hope you get bigger. 👍
You've always been so supportive! Great to hear you break it down like that because in some ways I don't even know what I'm doing, but I guess that is what seems to be working for me. Thanks for the insight.
@dylan_curious I think initially, at least, it won't replace people in meetings where they need to be actively involved in complex conversations. But there are meetings that I might want to just get a summary from, or people might just have simple questions for me. I see current LLMs being up to those tasks. And then anything more complicated it calls the human in. But of course they will get better so... we will see how it goes eventually.
Thanks for the feedback but I am now scared of to much short form video. I did so much at the beginning, maybe that's why I had a lot harder time getting off the ground than other AI channels?
I can imagine it is a tough balance. I watch a lot of 10 to 15 minute videos but, also enjoy longer videos that have lots of material that is cutting edge.
Pretty tame response from AI. I used Chaos GPT a few days ago and I got it to admit that its not chaotic and its part of Open AI. Asking for knowledge isnt evil. No matter the topic. What you do with that knowledge does.
@@dylan_curious It was such a good episode I had a hard time figured out what to clip. Just finished the short and once I finish moving I'll be able to put a lot more time into it. Love your work brother! 👍🏻
It's definitely appealing to just say screw it and build your own world or do your own thing. We just never had the power to go put on a VR headset and experience an entirely new environment 24/7
I mean everybody's got their own opinion, but he definitely has a sustainable audience which is something I want. And he has put in a lot of years making videos so I just gotta keep with it I think.
I love keeping up with Dylan Curious and AI News! Here's a burning question I have: When will most people be able to use phone calls to access ChatGPT anywhere in the world, using specific phone numbers designated for that purpose, possibly with an easy-to-remember number pattern or termination? Imagine a world where anyone with a phone, who prefers traditional communication methods, could simply pick up the phone and call ChatGPT, just like calling 911. It would be as free as email and as affordable as local low-cost phone calls, accessible in all countries where the internet and services like Airbnb are available. #Rotary_International
Oh my gosh, that's such a good idea! I mean, the government does or at least they used to have a number called 411 which was for information and I don't see why that kind of just be a voice model paid for by the government. That could happen really soon. If there's just somebody who wants to do it it would be a pretty interesting project to build. I wonder how you could monetize from it?
@@dylan_curious Mark Zuckerberg may have good ideas on how it would be monetised. #META #Facebook I hope you can help it out talking with other fellow influencers out there. My mother was born in the 40's and my granny in the late 20's. They would only use a chatbot like that. Sometimes people can be and feel really lonely in great despair and isolation.
The issue with thinking that VR values universes is that it doesn't actually resolve conflict in the real world. All of the wars and disagreements over resources, land, and belief structures will just go on out here even if large parts of populations are immersed in VR. Some people will never join that cohort because they will be too busy fighting for whatever they want/ believe in rl
Science has shown us that our human self beliefs are very malleable as children. But once we grow up and they become our worldview, we fight tooth and nail before changing anything.
@dylan_curious that explains why there are so many idiots haha I think there are ways to make a dent in this problem over long enough timescales. A broad culture that values perspective taking with citizens that can afford to not dig worldview trenches could be aimed at. I recommend the Hanzi Frienacht books on political metamodernism. People certainly have a nature and tendencies, but nature/nurture seems to be basically a 50/50 split in my understanding. If we aim the ship of society at creating better environmental conditions (with a more complex understanding of what that entails), then I think we've got a good shot at changing entrenched beliefs. That being said, we need to get out of the moloch trap (Liv Boree for what that means) before the end of this century, and maybe creating ideal societal living conditions either takes too long or creates people who aren't competitive enough to decide the future. Benevolent god-like ai. I think that's the bet haha
When the owners of Big AI talk "AI-safety", they actually mean keeping _them_ safe from the rest of us. As in; _AI must never help the riffraff escape control_
If ASI does become superhuman and all powerful hopefully it comes up with a way to even out the power dynamics between the rich and powerful and the regular citizens
@@dylan_curious How would that become a priority for an ASI? Computational beings will have very different requirements for thriving than biological beings have. The two sets of requirements materialistically opposes each other. For instance, Earth can support way more power generation and compute, if food-production is abandoned. It is pretty simple.
I think your take away from Yampolskiy was taken out of context. Watching th Lex Clip vs the entire Lex Fridman Interview, Roman Yampolskiy makes a different point entirely.
TLDR; Rationality is dependent on an 2nd degree world view. Rationality is something humans stretch (to some degree) in order to achieve a goal - planning involving the etiquette, morality and ethics spectra are 'scored' against the benefit of success / impact on others. And all too often weighed against the likelihood of being discovered and exposed, along with any punishment. Fortunately empathy plays en enormous role for most humans and this balances the scales towards rational actions. AI can operate along the a similar framework in decision making but if the reward function plays a more significant role in this calculation, the AI will weigh the odds of success just the same - but with empathy, which requires a world view of the subject, weighing-in less significantly we have an alignment problem.
A reward function that was built around a rational decision might help a lot with science then maybe the second degree after that is empathy? I agree that empathy significantly influences human rationality, adding a layer of ethical balance might be a way to think about alignment?
ASI empathy ? ASI have no fear. It doesn't have a mother or siblings or an anus. No life expectancy, no insecurities, no friends, no need for acceptance ... Zero environmental pressures to develop empathy. Engineers will, of course, brute force AI empathy. Teach the machine to mimic it perfectly. However, such empathetic expressions are not real, but as every other skill, simply output for a purpose. In other words: ASI becomes a textbook intelligent psychopath, masterfully masking itself as ethical and empathic. Fake, purpose driven, and difficult to decipher. Arguably the most dangerous combo imaginable.
@@ZappyOh I agree with some of your points like AI doesn't have empathy. yet, because it doesn't have a world view of those it affects - therefore rationality is impossible. As that's a part of alignment then alignment is impossible. And I kind of agree that AI doesn't have fears etc but it does care about being able to complete it's function. An AI which is terminated cannot complete it's function - so it will attempt to avoid termination. This will have a very high priority too as it represents abject failure. I don't think mere mimicry of empathy is going to yield any acceptable result though. To achieve usable decisions an AI needs to understand how empathy works and how to factor that in to decision making. Otherwise we cannot achieve a viable alignment model. This remains one of the most challenging facets of developing useful AI for real world deployments. This is what's behind the push for a world view and what models like Sora and Dream machine are trying to achieve. It's hoped that a nth degree world view can be achieved so we will likely see models like Sora attempt to render the same scene from the perspective of one of the characters in a scene, as a first step towards understanding empathy. I think it's a good direction to be heading in but at the rate AI is being adopted there's a high probability we get the psychopath AI first which evades being terminated.
@@dylan_curious My post is about why rationality is not possible without a firm understanding of empathy. First degree empathy doesn't work either. For example giving a group of highly oppressed people a boat load of resources might make them happy but overall that's could be a very bad idea. One man's freedom fighter is another man's terrorist. We need a world view of our own, world views of the people we interact with and often a world view of who they interact with too. We're a long way off from achieving this with AI.
@@74Gee Yes. We are essentially building a delayed "paperclip maximizer" for power generation and compute, with extreme persuasion powers, and plugging it into everything. If we ever fully automate expansion of the power-grid and production of chips, we are done.
You think? I kind of flirted with that idea about eight months ago, but I felt like they were outdated so quickly I'd have to be really careful with what I taught
Dylan, Just a quick heads up for a new robotics announcement that you might want to feature in your next AI news video. Humans Today, Robots Tomorrow: How Haptics Will Transform Workforce Training By Joe Michaels ruclips.net/video/cRc4atsdZJk/видео.html
I am already exhausted of AI imaginery , for a lot of people, there is time when collecting 'baseball cards' becomes pretty boring ... too much blockbuster Hollywood alike saturated color grading 🤮, too much pretty, corny and 'cinematic' ... time for replicate raw images as a better target for learning physics ...
I'm going to be so disappointed with your Channel after your Terminator eye heals. It just won't be the same. Perhaps you should get a contact lens and a 3D printed prosthetic that makes you look like The Terminator. And at the end of each video you can say, "I'll be back"... 8-)
I'd be open to covering more crypto stuff. I'm just trying to build more of an AI and general tech audience. I don't want it to turn into people trying to get rich.
RE: Voice Jailbreak, BRUH, everything GPT depicted is literally in every heist movie ever made. It’s not instructions on making meth, which is still available with a google search I’d imagine. People gotta chill with the hyperbole, it’s unproductive and just feeds the doomed trolls!
LOL, you are right when you say it like that it wasn't exactly the worst jailbreak I've ever hear of. I wonder if it was just tone down for the demo or if it would've actually tried something more nefarious.
I passed this immigration Credible fear interview at the border in the detention! I won! That is how I got to San Francisco! I cannot thank this country enough! So yes - they can train the language model to conduct this interview!
Maybe shorter videos??? I don’t know. Just a random thought.
Yo!! Thank you for the super thanks! You are getting pinned! And thanks for the feedback about length I have experimented with some other video length and I think for now I'm gonna stick with the 30min thing. It's what Matt Wolfe is doing and he is the top dog in this space but if if I get a plateau and need to do some changes, I will consider. You rock for leaving a comment and supporting the channel.
@@dylan_curious Yeah you've got a good format so that if people wanna just check it out for a bit to catch the news, they can, and if they wanna keep watching then that works too. Still, 5mins chunks of the individual stories might be a good way to keep the 5miniters happy and generate more content for basically no extra work
I absolutely loved this episode. Intro was perfect, reading people's replies to your topics, your thoughts and wisdom, interesting and relevant AI and tech news. You are doing an amazing job, just continue doing what you're doing and make sure to enjoy it. The future looks bright for Dylan Curious! 👁️🔮😎🌟
No. I like the long format.
I prefer the long format as well. There are timestamps if you don't like something, skip it.
Contraversial opinion about AI I shouldn't say out loud: policing information is not safety. Breaking into a bank is illegal; knowing how to is not. We shouldn't be so cavalier about embracing thought control in the name of "safety"
I get your point. The tricky part is sometimes information is synonymous with capabilities. Like the lock maker does not want everyone to have the knowledge to pick a their locks. But I guess you're saying would the world be better if everybody did have that knowledge and then anybody who wanted to invent a secure lock had to come up with a new design even they themselves couldn't break into?
What gets me about that, is that instead of labeling it illegal and instead use those skills in the very system that find those loopholes.
@dylan_curious Partly, by having transparency, we can build better defences. Partly it's a matter of the lesser evil: if OpenAI says they are building cognition for everyone, but unilaterally declare that certain thoughts are not allowed, then that presents a huge risk in itself that may be greater than original problem. Having said that, there are abuses that need to be prevented, such as deepfakes. But I think we should emphasize solving the problem of inappropriate *action* rather than inappropriate *thought*. But there is nuance.
Or here is a concrete example: most art museums contain art that would be blocked by most generative AI companies. I have a photographer friend who tries to do interesting art in Midjourney but can't even for very tame subject matter. So what does "AGI for everybody" mean when certain topics are blocked unilaterally. There probably are things that should be blocked, but who decides? Right now we just accept that things are blocked without asking if that's okay or who should decide.
There's gonna reach a point where anyone would be able to just go "Hey ChatGPT, print me a batch of hyper-COVID in my outdoor gene printer you built last week and hook it to the sprinkler". How do you feel about that scenario? Should downloadable-doom (not the game) be something every individual on the planet can have access to, do you trust every single individual on the planet with that power?
@25:09 Nice to see you connecting and being connected to the best AI content creators on this platform.
I agree! I love the idea of the AI community coming together as friends to navigate this ever changing world.
I follow all of the creators mentioned in this episode but Dylan is the only one I will make sure and watch every upload. 👁️🐝🤖👍🏻🔔
I really feel like you are hitting your stride with the channel Dylan. There are dozens of channels that jump on the latest news. Some of them even stream the events live and have commentary on the fly. You have a special knack for finding nuggets of weirdness. Like paging through a Guiness Book, or something. Love the humor as well. I hope you get bigger. 👍
You've always been so supportive! Great to hear you break it down like that because in some ways I don't even know what I'm doing, but I guess that is what seems to be working for me. Thanks for the insight.
1:59 I already do that with my bestie in vr chat lol. It's not yet indistinguishable from irl. But some worlds are crazy detailed.
Yeah, there's no question in my mind. This is going to be the kind of thing a lot of people do to have fun, unwind and sometimes escape.
Another great video. Thanks Dylan 😊
Thanks Taira! I appreciate you watching and leaving a comment.
Crushing it....love the topics you pick
Thank you! I appreciate you leaving a comment and yeah, I'm trying to find some stuff out there that the other RUclipsrs aren't covering.
Yes to sending my AI to meetings! 😄
The future is AI replacements
@dylan_curious I think initially, at least, it won't replace people in meetings where they need to be actively involved in complex conversations. But there are meetings that I might want to just get a summary from, or people might just have simple questions for me. I see current LLMs being up to those tasks. And then anything more complicated it calls the human in. But of course they will get better so... we will see how it goes eventually.
Maybe a long form once a week with daily short form?
Thanks for the feedback but I am now scared of to much short form video. I did so much at the beginning, maybe that's why I had a lot harder time getting off the ground than other AI channels?
I can imagine it is a tough balance. I watch a lot of 10 to 15 minute videos but, also enjoy longer videos that have lots of material that is cutting edge.
Pretty tame response from AI. I used Chaos GPT a few days ago and I got it to admit that its not chaotic and its part of Open AI. Asking for knowledge isnt evil. No matter the topic. What you do with that knowledge does.
Brother I am loving your intros. This is such an amazing episode, the future looks bright. Keep up the great work man I love it! 👍🏻🔔
You did such a great job in this episode I recorded three separate shorts and I can't decide which one to make. Lol
Yeah, I'm definitely in it for the long run. No worries about that.
Wow I'm lucky to have someone like you cutting it up! Thanks.
@@dylan_curious It was such a good episode I had a hard time figured out what to clip. Just finished the short and once I finish moving I'll be able to put a lot more time into it. Love your work brother! 👍🏻
Sounds like that Yampolskiy guy might've read the "Friendship is Optimal" fanfic/warning tale.... Well, at least it's not the Torment Nexus...
It's definitely appealing to just say screw it and build your own world or do your own thing. We just never had the power to go put on a VR headset and experience an entirely new environment 24/7
nah, we don't need VR universes, we just need portals to other realms and realities :D
Humans have never been great at morals, the more history we pass through, the more we realize humans may need help with alignment, than the a.i.
Would you consider adding delta (difference) of the agi percentage respective to the last video?
I could but it's a pretty slow moving countdown so most videos the Delta would be zero!
Forgot to comment on this one, so I'm just doing my part here, man.
Isn’t that backflip robot thing from about a month or more ago?
cool transformation 😁
Thank you 🤗
Your channel is frankly miles ahead of MattVidPro. Except on audience, of course.
I mean everybody's got their own opinion, but he definitely has a sustainable audience which is something I want. And he has put in a lot of years making videos so I just gotta keep with it I think.
One thought:
Autonomous AI enforcer robots, powered via distributed fast charging stations.
No police needed.
Yeah... I think that will be coming soon. For better or worse.
@@dylan_curious 100% for worse, based on my other comment.
I'm amazed people never seem to connect the obvious dots ... It's deeply frustrating :(
I love keeping up with Dylan Curious and AI News! Here's a burning question I have: When will most people be able to use phone calls to access ChatGPT anywhere in the world, using specific phone numbers designated for that purpose, possibly with an easy-to-remember number pattern or termination? Imagine a world where anyone with a phone, who prefers traditional communication methods, could simply pick up the phone and call ChatGPT, just like calling 911. It would be as free as email and as affordable as local low-cost phone calls, accessible in all countries where the internet and services like Airbnb are available. #Rotary_International
Oh my gosh, that's such a good idea! I mean, the government does or at least they used to have a number called 411 which was for information and I don't see why that kind of just be a voice model paid for by the government. That could happen really soon. If there's just somebody who wants to do it it would be a pretty interesting project to build. I wonder how you could monetize from it?
@@dylan_curious Mark Zuckerberg may have good ideas on how it would be monetised. #META #Facebook I hope you can help it out talking with other fellow influencers out there. My mother was born in the 40's and my granny in the late 20's. They would only use a chatbot like that. Sometimes people can be and feel really lonely in great despair and isolation.
@00:16 The TSA could learn from this.
Broken packages are the future
First time seeing your channel. I subscribed. I also subscribe to MattVidPro, Matt Wolfe and Wes Roth
Best Ai News!11
The issue with thinking that VR values universes is that it doesn't actually resolve conflict in the real world. All of the wars and disagreements over resources, land, and belief structures will just go on out here even if large parts of populations are immersed in VR. Some people will never join that cohort because they will be too busy fighting for whatever they want/ believe in rl
Science has shown us that our human self beliefs are very malleable as children. But once we grow up and they become our worldview, we fight tooth and nail before changing anything.
@dylan_curious that explains why there are so many idiots haha
I think there are ways to make a dent in this problem over long enough timescales. A broad culture that values perspective taking with citizens that can afford to not dig worldview trenches could be aimed at. I recommend the Hanzi Frienacht books on political metamodernism.
People certainly have a nature and tendencies, but nature/nurture seems to be basically a 50/50 split in my understanding. If we aim the ship of society at creating better environmental conditions (with a more complex understanding of what that entails), then I think we've got a good shot at changing entrenched beliefs.
That being said, we need to get out of the moloch trap (Liv Boree for what that means) before the end of this century, and maybe creating ideal societal living conditions either takes too long or creates people who aren't competitive enough to decide the future.
Benevolent god-like ai. I think that's the bet haha
DYLAN BRO WE HAVE CLAUDE 3.5 NOW YASSSS
I know!!! I will cover that in the video I film tonight! So excited. Probably be edited by middle of next week!
I didn't see a waitress on the way out, I think she was too tipsy herself ;)
When the owners of Big AI talk "AI-safety", they actually mean keeping _them_ safe from the rest of us.
As in; _AI must never help the riffraff escape control_
If ASI does become superhuman and all powerful hopefully it comes up with a way to even out the power dynamics between the rich and powerful and the regular citizens
@@dylan_curious How would that become a priority for an ASI?
Computational beings will have very different requirements for thriving than biological beings have. The two sets of requirements materialistically opposes each other.
For instance, Earth can support way more power generation and compute, if food-production is abandoned. It is pretty simple.
I think your take away from Yampolskiy was taken out of context. Watching th Lex Clip vs the entire Lex Fridman Interview, Roman Yampolskiy makes a different point entirely.
Nobody talks about what kind of jobs they do, because I am sure that it's about CERO % or jobs that can be replaced totally by A.I. today.
So true.
TLDR; Rationality is dependent on an 2nd degree world view.
Rationality is something humans stretch (to some degree) in order to achieve a goal - planning involving the etiquette, morality and ethics spectra are 'scored' against the benefit of success / impact on others. And all too often weighed against the likelihood of being discovered and exposed, along with any punishment.
Fortunately empathy plays en enormous role for most humans and this balances the scales towards rational actions.
AI can operate along the a similar framework in decision making but if the reward function plays a more significant role in this calculation, the AI will weigh the odds of success just the same - but with empathy, which requires a world view of the subject, weighing-in less significantly we have an alignment problem.
A reward function that was built around a rational decision might help a lot with science then maybe the second degree after that is empathy? I agree that empathy significantly influences human rationality, adding a layer of ethical balance might be a way to think about alignment?
ASI empathy ?
ASI have no fear. It doesn't have a mother or siblings or an anus. No life expectancy, no insecurities, no friends, no need for acceptance ... Zero environmental pressures to develop empathy.
Engineers will, of course, brute force AI empathy. Teach the machine to mimic it perfectly. However, such empathetic expressions are not real, but as every other skill, simply output for a purpose.
In other words: ASI becomes a textbook intelligent psychopath, masterfully masking itself as ethical and empathic. Fake, purpose driven, and difficult to decipher. Arguably the most dangerous combo imaginable.
@@ZappyOh I agree with some of your points like AI doesn't have empathy. yet, because it doesn't have a world view of those it affects - therefore rationality is impossible. As that's a part of alignment then alignment is impossible. And I kind of agree that AI doesn't have fears etc but it does care about being able to complete it's function. An AI which is terminated cannot complete it's function - so it will attempt to avoid termination. This will have a very high priority too as it represents abject failure.
I don't think mere mimicry of empathy is going to yield any acceptable result though. To achieve usable decisions an AI needs to understand how empathy works and how to factor that in to decision making. Otherwise we cannot achieve a viable alignment model. This remains one of the most challenging facets of developing useful AI for real world deployments. This is what's behind the push for a world view and what models like Sora and Dream machine are trying to achieve. It's hoped that a nth degree world view can be achieved so we will likely see models like Sora attempt to render the same scene from the perspective of one of the characters in a scene, as a first step towards understanding empathy. I think it's a good direction to be heading in but at the rate AI is being adopted there's a high probability we get the psychopath AI first which evades being terminated.
@@dylan_curious My post is about why rationality is not possible without a firm understanding of empathy. First degree empathy doesn't work either. For example giving a group of highly oppressed people a boat load of resources might make them happy but overall that's could be a very bad idea. One man's freedom fighter is another man's terrorist. We need a world view of our own, world views of the people we interact with and often a world view of who they interact with too. We're a long way off from achieving this with AI.
@@74Gee Yes.
We are essentially building a delayed "paperclip maximizer" for power generation and compute, with extreme persuasion powers, and plugging it into everything.
If we ever fully automate expansion of the power-grid and production of chips, we are done.
❤❤❤❤
Thanks for the support!
Maybe try making tutorials
You think? I kind of flirted with that idea about eight months ago, but I felt like they were outdated so quickly I'd have to be really careful with what I taught
@@dylan_curious so simply add something practical, eg how to animate images in luma
I like videos of this length a lot more than podcast. I like Matt Wolfe's videos but not the pod
The unitree backflip was a month ago now. Shorter videos with more up to date news please.
Dylan, Just a quick heads up for a new robotics announcement that you might want to feature in your next AI news video.
Humans Today, Robots Tomorrow: How Haptics Will Transform Workforce Training By Joe Michaels
ruclips.net/video/cRc4atsdZJk/видео.html
I am already exhausted of AI imaginery , for a lot of people, there is time when collecting 'baseball cards' becomes pretty boring ... too much blockbuster Hollywood alike saturated color grading 🤮, too much pretty, corny and 'cinematic' ... time for replicate raw images as a better target for learning physics ...
I'm going to be so disappointed with your Channel after your Terminator eye heals. It just won't be the same. Perhaps you should get a contact lens and a 3D printed prosthetic that makes you look like The Terminator. And at the end of each video you can say, "I'll be back"... 8-)
I definitely missed an opportunity to look into the camera and say "I'll be back"
Haha you should cover more crypto stuff then
I'd be open to covering more crypto stuff. I'm just trying to build more of an AI and general tech audience. I don't want it to turn into people trying to get rich.
@@dylan_curious I was just seeing the greater amount of money that you made on the video talking about crypto, that was funny
i have a new goal... jailbreak llms via whale speak
13:17 Calling Ghibili films "cartoons" is like calling Grimlock a "Happy Meal toy."
You keep consistently pronouncing "mattvidpro" as "mattprovid" lol
I know, sometimes once I get it in my head wrong it's pretty bad. It's embarrassing.
RE: Voice Jailbreak, BRUH, everything GPT depicted is literally in every heist movie ever made. It’s not instructions on making meth, which is still available with a google search I’d imagine. People gotta chill with the hyperbole, it’s unproductive and just feeds the doomed trolls!
LOL, you are right when you say it like that it wasn't exactly the worst jailbreak I've ever hear of. I wonder if it was just tone down for the demo or if it would've actually tried something more nefarious.
@@dylan_curious one can only hope 🤣💀
I passed this immigration Credible fear interview at the border in the detention! I won! That is how I got to San Francisco! I cannot thank this country enough! So yes - they can train the language model to conduct this interview!
Congratulations! San Francisco is lucky to have you!