00:21:12
◼►
I'm like a two docket of misunderstanding.
00:21:14
◼►
It came up at one point, Jonas took one computer science class in high school and they had to submit their, their, uh, read me files in markdown format.
00:21:24
◼►
And I said, you know, I said, you know, and he's, he just rolls his eyes.
00:36:33
◼►
You don't say last mile, but it's almost like, you know, the year of Linux on the desktop.
00:36:36
◼►
It always feels like we're about three months away from vehicles, but I, I think it felt like we were closer to that in some ways three to five years ago than right now, because as we, as that last mile ends up being a lot longer than a mile, exactly what you're describing the training that is involved in getting.
00:36:54
◼►
So this is really going to sound random, but I had a friend whose dad was a cop and was the head of the canine division in our Sheriff's department where I was from.
00:37:03
◼►
And anyway, long story short, he would say he loved Rottweilers and they had a Rottweiler as a pet and Rottweilers are so interesting.
00:37:09
◼►
And he, I don't know, he is, he's a cop and he liked dogs.
00:37:17
◼►
And if they think that you're not being a good leader, they will like you will, they will run the house, not because they're mean or because they're trying to annex the Sudetenland, but just because like they're not aggressive.
00:37:29
◼►
It's just, that's how their brain works is that they need that.
00:37:32
◼►
And the thing he would say, another thing he said that it really stuck with me that I still think about was like, you tell, so you can tell so much about a given, in this case, Sheriff's deputy based on how their dog acts.
00:37:42
◼►
That like, if it's not a very strong person with good character and who's got clear lines in their life, then that ends up being reflected in the dog.
00:37:50
◼►
But the thing he said to me that I still think about the most was something along the lines of the thing is a Rottweiler is like, it's almost like a Marine.
00:37:58
◼►
Like it's had the training and the background to know exactly how to follow orders flawlessly.
00:38:03
◼►
But he says like in the best ones, they also, they somehow seem to know when to rescind that order and do something else instead.
00:38:12
◼►
That to me is what separates AI from what you're describing here is if then things, because you could have an endless array and nested ways of doing all the defense stuff said the non-programmer, but like the idea of an AI, taking an AI and saying, Hey, here's something we're trying to accomplish.
00:38:27
◼►
And here's the basic rules of the road.
00:38:28
◼►
Like if you read like in the chess cheating thing, if you read about the ways that there are chess computers now that will do behaviors that are absolutely illogical, every human, and yet they win the games.
00:38:40
◼►
They are, it's the ultimate way of thinking outside the box.
00:38:42
◼►
And you get into this Kobayashi Maru thing where you start saying like, Oh, well, Hey, you're, you can't beat the test by cheating on the test.
00:38:49
◼►
And it's like, well, okay, well, you didn't put that in the instructions.
00:39:00
◼►
I, the thing that, where I get in over my head is with these, with the training that they do.
00:39:07
◼►
So they take the way that modern machine learning works and there is, and again, I'm a little over my head.
00:39:13
◼►
And again, I have a degree in computer science and I took, I think I took two courses in artificial intelligence back at old Drexel university.
00:39:20
◼►
I do believe there actually is a meaningful difference between AI is still the sort of broader parent umbrella for all of this.
00:39:28
◼►
And machine learning is not just, Oh, let's just give it a new name after 20 years to make it feel fresh.
00:39:36
◼►
It's it is a specific way of doing AI where you throw a gazillion examples into a corpus and train the model so that, and so you know that these are all images of.
00:39:52
◼►
Dogs, a hundred, a million images of dogs.
00:39:57
◼►
And you know that it's a million photos of dogs and you also have a million photos of cats and you throw them at the model and say, these are cats.
00:40:30
◼►
And then all of a sudden the model is actually very uncannily good at even identifying.
00:40:35
◼►
Isn't the killer part of this, the amazing and befuddling, the part that breaks my brain is the idea that it figures out what that route needs to be to get to where you're happy with the results.
00:40:46
◼►
That it's the, that's the machine learning part though, right?
00:40:49
◼►
Is it's not that we're there telling it what to do.
00:40:51
◼►
It's that we're saying more like this, less like this.
00:40:54
◼►
And like the chess robot that figures out these quackity ways to win at chess, it's thinking quote unquote in a way entirely different from the way that we would think.
00:41:18
◼►
And that's part of the, it's a little scary, you know, and I'm, I, you know, and I know that there are, you're in my entire lifetimes ever since 2001 in 1968.
00:41:30
◼►
There's an entire field of science fiction movies about AI systems gone wrong, right?
00:41:35
◼►
I mean, it's it, you know, you probably count on one hand, the number of science fiction movies,
00:41:40
◼►
like one of those Boston Dynamics projects where the robot rips the doorknob off or something like that, where it's like, well, you told me to open the door.
00:41:49
◼►
But it is a little scary, though, that even the machine learning experts who make these models that can do ever more amazing things really don't know how they work inside.
00:42:00
◼►
Like the internal logic of the system is impenetrable because if it were something that you could then it would spit out like a traditional computer program and you could read, then we'd be like, oh, I see.
00:42:13
◼►
Oh, we just missed this obvious thing.
00:42:27
◼►
And I'm not surprised that it's possible yet, but it still is amazing to see is it.
00:42:33
◼►
And it feels like pulling the idea inside out.
00:42:38
◼►
So me throwing a picture of a surprisingly dog like looking cat.
00:42:45
◼►
It is a cat, but damn if that cat doesn't look a little bit like sort of like a dog and throwing it at a computer and the computer says, Nope, that's a cat.
00:42:53
◼►
And it's like, ah, that's pretty good, but I'm not surprised.
00:42:56
◼►
But when you can say to a computer, draw me a picture of a cat that looks like Mike Pence, get off my hard drive in this, in the style of Leonardo da Vinci's notebooks.
00:43:07
◼►
And all of a sudden, all of a sudden it looks like Leonardo da Vinci was inventing a feline version of Mike Pence.
00:43:21
◼►
But you can see, though, why this has been an area of intense research for AI researchers for decades, because as a proof of this is a quote unquote intelligent system, giving a plain language text string.
00:43:35
◼►
Show me a picture of Donald Trump looking scared and alone in a large office with a can of Diet Coke in the style of Rembrandt.
00:43:45
◼►
Reminds me, yours reminds me of Edward Hopper a little bit.
00:43:47
◼►
The kind that looks like Nighthawks at the diner.
00:43:50
◼►
I'll try to remember to put some of these in the album art for the episode here so people can look down at your phone as you're listening to us and see some of these images that you and I have generated.
00:44:01
◼►
This one that I have of Trump with a can of soda looking scared and alone is unbelievable.
00:44:06
◼►
But you can just give a simple text string, not like some kind of, oh, you've written a little computer program and you need to know this crazy, this crazy, not crazy, but a complicated, precise, like learning to write SQL database queries code to do it.
00:44:27
◼►
Where you just you don't need to know a query language.
00:44:29
◼►
You just type a string and you wait and then you get output.
00:44:34
◼►
And then, and like with a Google image search, which the more time I spend with these things, I'm more amazed I am that Google has ever worked at all.
00:45:10
◼►
Even if you, I, I've gotten better based on looking at other people's results.
00:45:14
◼►
I've gotten better at realizing what kind of cues tend to be more or less effective, but like you really don't know what you're going to get.
00:45:33
◼►
But you don't, but like the point is like the, if you haven't played with this and that's fine, probably have like stuff to do.
00:45:38
◼►
But like if you've done this, it's bananas, the results that you come up with and it's, you can kind of go like, oh yeah, I can, I guess I kind of see how you got that.
00:45:46
◼►
But part of the reason I love it so much is I have no idea what's going to come up, what it will look like.
00:45:53
◼►
And could I even say, was this one good?
00:47:23
◼►
They always do these seasonal recipes.
00:47:25
◼►
They also have so many options for stuff like you want vegan options or if you have allergies, you want to avoid certain ingredients, that sort of thing.
00:47:35
◼►
It's so many things to choose from every week.
00:47:37
◼►
You pick what you like and that's what you get.
00:47:44
◼►
I love, especially love the pre-portioned ingredients so that if you need something, you need celery for the thing, they send you enough celery.
00:47:53
◼►
I like that they send me things that I would never choose or make myself.
00:47:59
◼►
But to me, that's part of the fun is like, these are dishes, sometimes they're the ingredients I go, but like, I like the fact that's nothing I would have like clipped out of a magazine myself.
00:49:38
◼►
I go to places like Las Vegas and buy nail clippers and things like that.
00:49:40
◼►
But like it's, there is this funny thing that, so I'm sitting here with this studio Mac, which is crazy powerful.
00:49:45
◼►
And when I'm doing this at home on the laptop, it's not quite as fun, but when I'm, even if I'm doing it at the office with everything else quit.
00:49:51
◼►
Cause this thing is really hungry for Ram and I guess processing, but what is it?
00:49:56
◼►
But it goes, it's just enough time for me to go, Oh boy, Oh boy, Oh boy.
00:50:07
◼►
And then something more salient for me is the whole, like, Oh mom, can I just stay for one more music video?
00:50:14
◼►
Because it might be standing delivered by Adam and the ads or like whatever, like my favorite video might come on and you went one more, one more, one more.
00:50:21
◼►
And like, I have to peel myself away from this app because I get so, and then I'll go like, Oh, brainstorm.
00:50:46
◼►
So stable diffusion is one of these systems, but as opposed to the other ones, you don't just go to them and they keep it all.
00:50:54
◼►
It's an open source thing and an open source corpus, and you can download it all and you can run.
00:50:59
◼►
It's the first one of these dinguses, these AI image generation things that you can run on your own computer as opposed to running on their server farm somewhere.
00:51:11
◼►
And apparently it was, so kudos to the project for making it, shipping it, putting it on GitHub and you can download it.
00:51:20
◼►
But apparently, and it's getting just what it's worth.
00:51:22
◼►
This one is getting upgrades and improvements at quite a pace.
00:51:26
◼►
Well, yeah, I noticed after a couple of weeks away from it that the results are already better, but.
00:51:31
◼►
Well, they've added you've got history now, which is huge.
00:51:34
◼►
And the other one is now you can do image to image.
00:51:38
◼►
So the price, the parent project is stable diffusion and diffusion B.
00:51:44
◼►
Like the insect is a project by Devan Gupta, who took stable diffusion, took the open source license and packaged it up into just a regular simple Mac app.
00:52:15
◼►
It's no different than like the complexity of installing BB edit, right?
00:52:20
◼►
You just download it and you drag it to your applications folder.
00:52:23
◼►
And it's the only difference is it's got three gigabytes of data to download.
00:52:27
◼►
And then all of a sudden you've got a regular nice looking Mac app with a text field where you just type whatever it is you'd like to get a picture of and you hit return.
00:52:39
◼►
And that this is the part that to me is a real interesting and also a real throwback to the early days of me using personal computers, which is in the early days of using personal computers.
00:52:53
◼►
The most fun stuff was the cutting edge stuff, and it always took forever, right?
00:53:05
◼►
It had a very similar feeling of like, well, this has been described to me in the following way.
00:53:11
◼►
And then when you wait for the tucka tucka tucka tucka tucka 9600 baud modem to like draw it for you.
00:53:16
◼►
Well, I remember it like the early 90s, like when I was at college and it was still a novelty that the Macs now had full color displays, right?
00:53:27
◼►
It's like, you know, this is the Mac was the black and white computer.
00:53:30
◼►
And now they had color displays and I had a Mac LC with a color display.
00:53:34
◼►
And then there this new image format came out called JPEG and JPEG didn't look like a gif file or at all.
00:53:45
◼►
They looked like photographs and you could just download them and open them.
00:53:50
◼►
But even once you had it downloaded, you could watch a JPEG sort of come into your screen from the top left to the bottom right as it would fill it.
00:54:00
◼►
And remember, there's the progressive JPEG.
00:54:05
◼►
So the younger made this file larger by putting crappy versions that'll come in first.
00:54:10
◼►
But so you got like you got like the version that you haven't put your glasses on yet first and then as just so it would fill the screen and then the higher res version would fill in.
00:54:21
◼►
Progressive JPEG was actually pretty clever because it did make had to make the file slightly bigger, but only slightly.
00:54:30
◼►
It felt faster because you got something that filled the full space that the image would eventually fill out roughly just a rough blurry version of it.
00:56:00
◼►
But anyway, once you could play video on your computer, it was amazing.
00:56:04
◼►
And like you said, but even before you could play this tiny postage stamp, low frame rate, high compression, low color palette video, you had to download it.
00:57:11
◼►
There was a guy named Peter Sishel as S I C H E L who had a thing for the Mac that would let one Mac be connected to the modem, but would share the connection with over the local network with others.
00:57:28
◼►
I forget what it was called, but I'll try to remember to put it in the show notes, but it was a wonderful piece of shareware.
00:57:33
◼►
But it was like we could share one modem connection and she could initiate it on her own from her iMac so that I didn't have she didn't have to go to my computer if I wasn't around.
00:57:43
◼►
As long as you remember to throw your TCP IP perhaps occasionally.
00:57:46
◼►
But it was so and that's what these the way that everything was so slow back.
00:57:51
◼►
Remember when you first got Photoshop and at this point just a few years later, all of a sudden it wasn't a big deal to open a photograph in full color on your Mac.
00:58:12
◼►
I don't want to say was illustrated three, but one of the there was an illustrator circa ninety two ninety three, something like around that time where you could have it in two modes where it was almost like.
00:58:22
◼►
Oh yeah, the best bezier curves or whatever.
00:58:25
◼►
And then when you wanted to see what it's actually going to look like, even let's just say like John, am I exaggerating if you want to see how your gradient turned out, like go get a coffee.
00:59:03
◼►
I remember when you'd print a complex page layout from Quark Express, just translating it all to postscript and sending it over to my image writer.
00:59:15
◼►
Finally, I would always print out in draft and it would be all jaggy because I was usually using New York and they would of course will look amazing once I did it on the laser printer.
00:59:31
◼►
Got to wait a second to see the cool thing.
00:59:33
◼►
Well, like no more than a second, though, like a minute or 30 seconds, I don't know.
00:59:37
◼►
And then you'd realize now and it's the exact same feel to me where you're like, oh, it's the same thing with like Napster where you're looking and we would just play the game like, oh, what about so and so?
00:59:49
◼►
There's an out there's music we haven't heard since we were in the mid 80s.
01:00:15
◼►
But these AI general image and then you realize you type slightly the wrong string or something and you're like, that's not that I thought that would get me something in the style of Norman Rockwell.
01:00:26
◼►
But it's like, oh, I see you instead of saying style of Norman Rockwell, you say illustrated by Norman Rockwell.
01:00:33
◼►
And you all of a sudden it looks like a Norman Rockwell.
01:00:35
◼►
I don't you know, I might be no, no, no, I it does.
01:00:38
◼►
Well, it's again, it's difficult to know we don't have a I don't have a baseline.
01:00:43
◼►
So I don't have a way of saying if I just enter in Mike Pence and if it's not clear, I'm obsessed with the way Mike Pence looks in a generated art.
01:00:49
◼►
I'm utterly obsessed with it because he's a very comical character to me.
01:00:52
◼►
And it's funny to see him with three breasts and four arms.
01:00:55
◼►
But but but yeah, yeah, changing the order of things changing that I discovered this when I went to another one of those sites.
01:01:23
◼►
In fact, an extremely detailed one that looks like SEO jamming.
01:01:28
◼►
Remember, you know, when you try and cram keywords like and make them FFF so nobody could see them like all those cheesy ways of trying to leverage SEO, putting a ton of stuff in there might take just about as long and you may or may not get a better result.
01:01:45
◼►
But you know, I'm involved because as much as I seem like a monkey hitting a button, the changes in what I choose for the prompt end up having an impact on it, which is exactly the kind of lizard brain crap that totally appeals to me and a lot of other people.
01:02:16
◼►
It is very satisfying to know it's running on your computer.
01:02:20
◼►
It is getting better at a noticeable clip.
01:02:23
◼►
I would say it is remarkably better one month later, which is not so it's one of these things that just turns on a part of my brain that starts pumping out.
01:02:44
◼►
Unfortunately, I don't have a lot of friends, so I don't have to explain why I'm spending all the time on this.
01:02:47
◼►
But if I were going to try to explain to somebody I and I've said this, I think the Syracuse I was saying, like, part of what I find so appealing about this is how I don't see how bad it is, but how weird it is.
01:02:54
◼►
And that there will be something there that my my brain, my cognition can recognize is a certain kind of thing.
01:03:01
◼►
But again, if it has an extra arm or something, you get into this weird, almost like at least I don't know how my dreams are.
01:03:08
◼►
I don't have dreams that are easy to tell as a story because it can be very visual, but it's very based in feelings.
01:03:13
◼►
And of course, nobody likes being told a dream.
01:03:15
◼►
But the kind of imagery, the closest thing I could think of, I think I mentioned this to you, is reminds me of the paintings of Francis Bacon.
01:03:21
◼►
So if you look up, what's the famous one, Francis Bacon, there's this one called Painting 1946, where it kind of looks like a side of beef dressed as the pope.
01:03:29
◼►
Like the stuff that guy was doing for paintings.
01:03:45
◼►
You're not going to reverse engineer this from somebody else's art.
01:03:47
◼►
Like if you go in and enter a prompt and get 10 results from it, it could be shocking how different and how extremely weird some of them are.
01:03:58
◼►
And then the contra that, then one out of every 15, you'll be like, wow, that's really normal looking.
01:04:03
◼►
Like I did one, there's a, oh, again, Mike Pence, but I was doing funny bits involving the X-Men.
01:04:09
◼►
And there's an artist I like a lot who's on Twitter, whose name is Bill Sienkiewicz.
01:04:13
◼►
He did stuff like Daredevil and Electra and stuff.
01:04:34
◼►
It's not just when Mike Pence and Cthulhu have, have an odd number of arms.
01:04:39
◼►
But it is, there is a very interesting dream logic.
01:04:42
◼►
And I think this is part of, without getting too far into the, for me anyway, without getting too far into the rat hole of like, what does this mean for the future of art?
01:04:50
◼►
There's something weirdly stimulating about it.
01:04:52
◼►
Even if this is not going to go create a logo.
01:05:26
◼►
Cause, uh, Donald Trump, comma graffiti, comma black and white comma style of banks, style of Banksy.
01:05:34
◼►
These are two that I got that I just loved.
01:05:36
◼►
I especially love the bottom one where it, I'll try to put this one in the,
01:05:42
◼►
cause sometimes it will see, you'll see that letters, but we, you can recognize something as, Oh, whatever this is based on had letters in it, but it'll be like, you know, almost like the teachers in a peanuts cartoon.
01:05:51
◼►
My one, one, one, you'll see something that looks like this is the closest to like the real letters that I've ever seen.
01:06:30
◼►
Memberful allows anybody, you listening, not you Merlin, but anybody out there listening out in podcast land, you can build a sustainable recurring revenue through memberful.
01:06:40
◼►
It is the easiest way to sell memberships to your audience, and it is used by some of the biggest and best.
01:07:24
◼►
I've gotten enough of my money, you know, they got their nut.
01:07:27
◼►
Whoever the hell owns Paramount plus doesn't really need more of my money.
01:07:32
◼►
They've gotten a lot of it over the years, but the shine heart week company, my favorite stuff that I consume.
01:07:37
◼►
When I look back at my weeks media consumption, my favorite stuff comes from the stuff that I remember and the stuff that I feel was actually nourishing for my mind comes from independent producers.
01:07:49
◼►
And even if it's just entertainment, it's still, it feels more, it's what I want to support and the way you can support it is through memberful.
01:07:56
◼►
And if you're the person type of person who makes independent media out there, you can use memberful to offer memberships to your audience.
01:08:04
◼►
You can get started for free with no credit card required.
01:08:08
◼►
I'm I subscribed to a bunch of memberful sites.
01:08:14
◼►
Like we, that's what we use the podcast network where Sir Q said, and I have a show. I use it to, I'm pretty sure of all of these, um, six colors and you know, Jason and Dan and that, that team and Matthew Casanelli.
01:08:38
◼►
I don't want my own breaks and I don't handle my own money stuff.
01:08:42
◼►
Like that's, I'm so happy to have somebody else.
01:08:45
◼►
John, these kids today, they're not going to remember that around the time that you and Amy were sucking down all those Tom Petty bootlegs in your, in your room.
01:08:52
◼►
Like it, it used to be of this is, this is that reason that terrible man runs the car company now is because it used to be hard to do money things on the internet.
01:09:01
◼►
And to this day, I mean, just the stuff you've got to deal with.
01:09:05
◼►
Think about when you sell t-shirts somewhere and you got to deal with the vat and all of that.
01:09:09
◼►
Having somebody else handle that stuff and then on top of it, have these it's value added services, things that you can do through memberful.
01:09:17
◼►
I'm not, I'm not here to stand for memberful.
01:09:21
◼►
If you want to, if you want to, if you're making a thing and I don't use the word content, I don't use that word, but if you're making stuff and you want a way to put a cash register on it, check them out.
01:11:26
◼►
I thought I was just signing up to get access to mid journey and I figured that I would be.
01:11:30
◼►
That's a pretty I've run into that a handful of times where to like get in on like to get a beta of something or whatever you do it like through a discord.
01:11:38
◼►
Yeah, that's well, it turns out that there is no like mid journey website where you get a credential like a username and password and you type things in their website and then the website gives you images.
01:11:56
◼►
And so what happens is they just give you access to their discord and you there are these channels in the discord for newbies.
01:12:04
◼►
I don't know what they call them, but you're a new user.
01:12:06
◼►
You haven't paid for anything and you get into one of these channels and you just start typing like commands and they're relatively simple.
01:12:17
◼►
It's a little bit like a command line interface where it's like you say like I'm getting get the syntax wrong, but like using a command line interface.
01:12:24
◼►
In the sense of that, there's like attributes like there's verbs and nouns and like flags.
01:12:31
◼►
Yeah, well, like it'd be like a new image and then you just type like these other ones new image Donald Trump eating a mouse and then you hit return.
01:12:39
◼►
Yeah, and it takes again, it still takes like 30 seconds for your things to come back and it comes back on mid journey with a like a little four square.
01:12:51
◼►
1234 with four versions of which might vary greatly of what your prompt was, and then you can pick one of them like, oh, number two.
01:13:01
◼►
I like and I forget what the command is, but it might be like M2 and M2 would say take number two and do another revision on that and give you another more like this.
01:13:12
◼►
And then when there's one that you really like, there's you just say like, oh, that's the one give me a higher resolution of that one.
01:13:17
◼►
And then it gives you a higher resolution of that one. And each one takes I don't maybe 30 seconds.
01:13:23
◼►
I here's where it blows your mind. This is where it is absolutely bananas.
01:13:28
◼►
And I might this might be the last episode of my podcast where I have my full listenership because some portion of the people out there will I you're going to be if you've never tried to lose them, they're going to disappear into mid journey.
01:13:44
◼►
Because while you're doing this, you're in a public discord channel with people from all over the world doing the same thing in your channel.
01:13:52
◼►
So as a free user of mid journey, whatever it is you're making is not private. It is on their channel. You can pay them and subscribe.
01:14:02
◼►
It's like you pay like 200 bucks a year. I don't know something like that. And then you can get a private access to mid journey where you can type things that nobody else can see in their discord.
01:14:11
◼►
But while you're learning it, you're in a public channel. And it's not just one channel with all 40 million people using it around the world.
01:14:19
◼►
They somehow automatically keep only I don't know, a couple dozen people at a time in one of these channels.
01:14:26
◼►
But you see what they're making and what they're typing. And these are people who are all way better at it than me.
01:14:34
◼►
Right? Because I've never been here before.
01:14:37
◼►
These they know, like they know, like the incantations.
01:14:40
◼►
They know the incantations and you can see their incantations as they type them.
01:14:45
◼►
And then so while you're waiting 30 seconds for your incantation to give you your results, you're watching theirs flow by in the channel.
01:14:56
◼►
Oh my gosh, and discord is it turns out to be wonderful for this.
01:15:00
◼►
It's the first it's discord went from a thing that I never wanted to use and never enjoyed.
01:15:05
◼►
God bless all the other sites out there that have discords.
01:15:07
◼►
I know some people like it, but I find it my brain is too simple and too easily confused.
01:15:12
◼►
It's an intuitive place to do something.
01:15:14
◼►
But while you're waiting for your results, you see these amazing results from other people and occasionally unamazing results.
01:15:20
◼►
But you also got to see and you can scroll back to see what they typed to make it.
01:15:39
◼►
And while you're waiting, all these other results are streaming by.
01:15:44
◼►
And then next, next thing you know, it's tomorrow and your wife, your wife is like, where were you?
01:15:50
◼►
And you're like, ah, but if this were gambling, I'm doing, um, whatever slots and you're playing craps like you're playing a community game.
01:16:00
◼►
You know, I love the gambling, but I'm going to give you a better analogy.
01:16:03
◼►
To me, it is like discovering coin op video game arcades when we were young, where it is because it's there's the money is incidental, right?
01:16:18
◼►
You can suddenly lose your money, whereas in an arcade, you kind of spend your quarters at an even clip.
01:16:22
◼►
But you let unless you're playing the Star Wars sit down game, they're all going to be a quarter.
01:16:26
◼►
But like when you would discover like the first time, maybe you're on vacation, maybe a new arcade opened at your mall.
01:16:32
◼►
But like the first time you found a new arcade and you could tell it was a good one because it was big and they had not just a game you'd never seen before, but multiple games you had never even seen before.
01:16:46
◼►
You wanted to play Donkey Kong, but somebody who's really good is already there and they've already got a bunch of quarters up on the thing.
01:16:52
◼►
So, you know, they're going to be there a while.
01:16:54
◼►
So you just walk the aisle and you can watch other people play games.
01:16:58
◼►
And in the same way that maybe you'd see a game that you had tried before, but you were never very good at it.
01:17:05
◼►
And oh, there's, here's a girl who's playing Galaga and she's really good.
01:17:12
◼►
Oh, I never thought to move around like that.
01:17:15
◼►
I was like, oh, the asteroids watching someone else play asteroids.
01:17:19
◼►
I was like, oh, it had not even occurred or in that matter defender like games where you're like, I had no idea that that was a thing that I could even do.
01:17:26
◼►
Defender was an incredibly difficult game.
01:17:30
◼►
Well, and of course Stargate, but like you would learn, like hang out in this area, wait for this thing to happen, go into the Stargate, do this thing.
01:17:41
◼►
I enjoyed the novelty of the track ball and it had really cool sounds, but I would die very quickly and I didn't want to squander my quarters.
01:17:49
◼►
And then I'd see somebody who was good at it and say, oh, you can kind of hide there when that spider drops down.
01:17:55
◼►
And you're once your, the spider is only going to go that direction.
01:17:58
◼►
So once you get past it, you can, yeah, I get it.
01:18:00
◼►
And that's what it's like hanging out in this mid journey discord where except there are no quarters.
01:18:55
◼►
I didn't see anything that I wouldn't have wanted a teenager to see.
01:18:59
◼►
I don't, and I don't know how much of that is that their training model blocks stuff that would be violent or terribly sexual or something like that.
01:19:28
◼►
And like, well, you know, we've asked that about a lot of stuff in the past, but, and sometimes some of us get a nose for going, like, I don't know what this, like, I don't know how this becomes the next job.
01:19:38
◼►
I didn't know how it would become a job in 1993 to make web page or 94 to make web pages.
01:19:43
◼►
And somehow it did kind of, you know what I mean?
01:19:44
◼►
Like, but it's, I think it's going to be crazy interesting and we don't have anywhere near time to talk about it today, but what, there's so many things where I used to look down my nose and go like, Oh God, really?
01:19:54
◼►
Like you're worried about robots rights and you think all these things and like, but more and more I will, once I actually learn something about something like this, I go, Oh, now I understand what the big deal is, but also I understand why this is a complicated thing.
01:20:06
◼►
It is all the people, and again, they talked about this on ATP, but all the people on who's, if you're in the corpus, like you may be like, Hey, like you can't use my, my drawing to do that or whatever.
01:20:17
◼►
You're trying to put me out of business.
01:20:18
◼►
And as everyone, we try to predict the future.
01:20:20
◼►
I think like, we don't even know what we don't know yet about this.
01:20:23
◼►
That's, that could be a whole episode.
01:20:26
◼►
I thought about that whole can of worms of the ethics.
01:20:28
◼►
I agree. And Charlie Warzel illustrated a piece in his galaxy brain newsletter a month or two ago, probably closer to two months at this point, where his hero art for this issue of the newsletter was an AI generated picture of Alec, dipshit, info wars.
01:20:49
◼►
And his idea was, look, the rules of his newsletter at the Atlantic is that there's got to be a hero art thing at the top, and they have an access to Getty images, and he can type out a hero art, if I understand correctly, john is talking about when you have a primary large image on a page that will then also be used in stuff like a card, right? Twitter, right.
01:21:06
◼►
And yeah, almost every website, other than mine does it with every post or sites.
01:21:12
◼►
It's not about Safari tabs, you're not interested.
01:21:14
◼►
But his thinking was, I don't want to just get one of these stock images from Getty photos from Getty images of Alec Jones, which have all we've all been used before.
01:21:24
◼►
And his budget wasn't such for each issue, you know, like you pay, I don't know what they pay Getty for that 150 bucks or something like that, or whatever for a licensed photo.
01:21:33
◼►
He doesn't have the $1,000 it would take to commission a human artist to draw an original piece of art that wasn't even an option.
01:21:42
◼►
And people freaked out that him using AI generated art, and it was really good.
01:21:47
◼►
It was really the tone of it was amazing.
01:21:49
◼►
And he took it to heart and followed up and took it all on the chin and said, Look, I wasn't really thinking like that, but I'm not going to do it again.
01:21:57
◼►
And his over the objections were that was it the whole you're going to put artists out of exactly more or less to encapsulate a complex argument into something?
01:22:04
◼►
Yeah, I was not I'm not trying to be reductive.
01:22:07
◼►
But I think that saying I'm not going to use that I won't do this again is so is going to look so foolish because everybody's going to use this.
01:22:14
◼►
And it's if you're an illustrator, I don't think you're going to be put out of business.
01:22:17
◼►
But you know, I think human illustrators have going to have to twist their styles into a direction that the AI systems can't replicate, right.
01:22:26
◼►
And so it there's will quickly I think we'll get to a point where everybody will recognize the AI generated ones no matter how original they are.
01:22:40
◼►
We could talk about how it's wrong and it's putting musical artists needed to get paid and blah, blah, blah.
01:22:45
◼►
And but it's you can't put it back in or we could talk about how Spotify and Apple Music don't pay artists enough from this the streaming the 10 bucks a month you pay to your service of choice.
01:22:55
◼►
Not enough of that music goes to the artists that might be in the news and the newspaper where I was a clerk in 1988 doesn't sell as many copies in right.
01:23:07
◼►
You've been that you've what was your you have those stories about going the paper used to work out where you go in the morning and you could see all the like the copies being sent out.
01:23:14
◼►
And I'm not trying to say I'm not trying to be unkind at all because you know, we all have to contend with all of these things.
01:23:19
◼►
But a lot of times there are opportunities to it's just that we can't you're not a Toyota plant.
01:23:26
◼►
You can't be just retooled into like, oh, now I make stuff that's better than AI like I don't know what the answer to that is, but I do know that it's probably a pretty good chance.
01:23:39
◼►
And there is the question of the training right?
01:23:41
◼►
And if these millions of model of images that went in to feed the models, how are those people being compensated?
01:23:46
◼►
I guess they're like a doesn't seem like I'm you you'll remember along with me that in the early days of the web like you think about when like photographers first started putting up their portfolios.
01:23:55
◼►
And within about a week, we're like, forget that.
01:23:58
◼►
And so you get the like the wedding photographers style like watermarks, everything.
01:24:59
◼►
I know we have a limited amount of time, but now there is a part of me that thinks you're just trying to ditch me because you want to go get back in the diffusion.
01:26:27
◼►
And when you're ready to launch, just use that same offer code talk show, save 10% off your first purchase, including prepaying for up to a year, save 10% for up to a year.
01:26:36
◼►
And again, that 30 day period, your website is real.
01:32:27
◼►
That's not the way to solve this problem.
01:32:29
◼►
When I was in college, we had to eventually go to something that, this will probably go in there at some point, we called the one fork rule, which is like when we reached a certain point of three 20 year old men living in a room, living in a house together, we had to like call lockdown, basically, rank rank, all the clock songs are going off because we're going into one fork mode for the next week.
01:32:46
◼►
And so we can earn the ability to have more than one fork without leaving a mess.
01:32:51
◼►
A lot of it is just stuff that like seems intuitive. That's not the stuff in your life where you need to be kind of shaken out of, I mean, stuff even from the last few years, call somebody what they want to be called and don't be a dick about it.
01:33:33
◼►
Well, I was gonna say was that with the 43 folders days, what I realized too late when I was already into doing this book project was that I have a certain sweet spot for certain amounts of words.
01:33:43
◼►
Like I think I've done a pretty, I've done pretty good in the past with 140 and occasionally 280 characters, but that's why Twitter was fun.
01:33:53
◼►
And then, but then like honestly, what I discovered was that I, the stuff that I'm most happy about having written and that I like sharing with people, I don't know, 3000 to 5000 words.
01:34:05
◼►
I don't have a brain that works in 12,000 word increments.
01:34:09
◼►
And that's what puts me into that spin or like why I need something like mark down to keep me focused because then I'll be doing all I should make this an outline.
01:34:17
◼►
I should make this a mind map and all these different things that are taking me further and further away from just typing words.
01:34:23
◼►
I mean, like I thank you for saying that because I really appreciate it.
01:34:28
◼►
Like any truly like any great work of art, it best just enjoyed for what it is.
01:34:35
◼►
So what I, what I, what I imagine, rather than talk and talk about it and read my favorite ones or whatever, I just wanted to encourage everybody listening to just go read it.
01:34:53
◼►
It'll be, it'll, what I eventually want to get to someday is to make it where you could say, show me all the stuff about this kind of thing, that kind of thing.
01:35:34
◼►
But you know what I like to do as I like to just sit back with something in an iPad and just sit there and scroll and have it be the nice readable size font.
01:35:43
◼►
Well, have me on someday to really talk about this because like I grew up, I cut my teeth on reference books.
01:35:49
◼►
I loved things like obviously encyclopedias, but I also love things like the book of lists and the book rules of thumb and all these different things where there's just so many like little tidbits, little heuristics.
01:36:01
◼►
For like trying to make it through the day.
01:36:03
◼►
And anyway, I appreciate you saying that.
01:36:06
◼►
Have me back on another time and maybe it'll be more, there'll be more stuff there, but it'll still be in Markdown because I, that's, that is the way and I love you for making more.