Ditching Hourly

Jonathan (00:01.626)
Hello and welcome to Ditching Hourly. I'm Jonathan Stark. Today I am joined by guest Erin Austin to talk about the legal and ethical issues of AI. Erin, welcome to the show.

Erin Austin (00:12.35)
Thank you for having me. Glad to be here.

Jonathan (00:16.026)
Awesome, thanks for joining. So could you tell folks who maybe haven't come across your name before a little bit about who you are and what you do?

Erin Austin (00:22.942)
Sure. Yeah. I'm Erin Austin. I'm founder of Think Beyond IP. That is a law firm that focuses on helping experts turn their expertise into intellectual property assets so they can decouple their income from their time. And so I talk a lot about copyright issues as well as AI as a matter of fact, and contracts that affect both.

Jonathan (00:39.322)
Ooh.

Jonathan (00:53.338)
Interesting. So you sent out an email, I think it was a couple days ago, was it email or LinkedIn? I think it was an email.

Erin Austin (00:59.646)
It was both, honestly.

Jonathan (01:02.202)
I've spotted. And it was about the difference between plagiarism and copyright infringement, I think. Is that right?

Erin Austin (01:12.126)
Yeah, it was in response to a LinkedIn post where someone, I don't know how it was. I know somebody actually copied me into the discussion where someone was asking, do you think using AI and not crediting the AI platform or that it was written by AI is that plagiarism? And there are a number of comments about it. And I jumped in.

Jonathan (01:34.394)
Hmm.

Erin Austin (01:38.942)
to say that I did not think that it is plagiarism and that is because I look at plagiarism from an ethical point of view and so that was the jumping off point for the newsletter and that I wrote about that issue.

Jonathan (01:57.37)
Interesting. That's so cool. Okay, so how would you define those terms? I mean, is that something that we can define on our own? I guess, ethically speaking, that would be the case, right?

Erin Austin (02:08.582)
Yeah, I mean, so plagiarism. So that is the ethical breach of stealing somebody else's ideas. And so you can have the legal issues with copyright infringement. Well, let's just talk about the ethical issues of stealing someone's ideas. So we're all familiar with, you know, from our school days that we're supposed to quote, you know, cite our sources.

and that we took something from the encyclopedia, which at the time, they were still encyclopedias when I was writing school papers, and we didn't quote them, then, or cite them, then we were plagiarizing. And so we are ethically obligated to give credit to where an idea comes from. And so...

Jonathan (03:01.498)
Okay.

Erin Austin (03:03.518)
you know, that can have implications, obviously, in the school context. If you plagiarize and don't get credit, you can, you know, get a failing grade or even be kicked out of school. In a professional context, if we are taking somebody else's ideas as our own and not giving proper credit for it, then we can be accused of plagiarism there. And but ultimately, like,

Who does that ethical obligation run to? And so the question was, you know, is it plagiarism not to credit AI? And in my opinion is that we do not have ethical obligations to AI. Like it is not the creator of any ideas. It can take a bunch of data points, you know,

Jonathan (03:37.434)
you

Erin Austin (04:02.43)
billions of data points and do its magic that it does and have an output. But is that AI's idea? And I would say that it is not. And there's two sides of the coin to plagiarism. One, making sure the originator of the idea is credited. And, you know, I don't think that AI needs that.

Jonathan (04:13.754)
you

Erin Austin (04:32.446)
and as the plagiarist to make sure that the plagiarist isn't getting credit for the big idea that is not his own. And when we look at the output as of today, the type of output that we're getting from AI is that kind of reputation making output. And so if, you know, again, we're just taking something that gets spit out of a

AI, an agenda AI platform, and we put our name on it. Are we getting something that would be considered thought leadership? Are we getting something that that people would go, well, you know, that's really, you know, and my experience is that we aren't. And so from both sides of the coin, I feel like we fail the plagiarism analysis.

Jonathan (05:04.058)
Mm -hmm.

Jonathan (05:15.898)
Yeah.

Jonathan (05:29.146)
Okay, so if I can, to make sure I'm understanding that, I want to summarize it back to you. If plagiarism is stealing someone else's ideas, you'd say, well, it's not someone's idea that AI is a machine. It's like stealing the answer from a calculator. And the other thing is the ideas just aren't that good. Well, if they're not ideas, the output's just not that good.

Erin Austin (05:45.534)
Perfect.

Jonathan (05:57.37)
So it's not really gaining the air quotes author, the person who used the AI to generate the output. It's not really getting them anything.

Erin Austin (06:07.518)
Yeah, I mean, what are you gaining reputation wise when you are using AI generated content as your own? Like what is the upside in building your reputation? And I mean, there could be some, I mean, I imagine there's some people who have figured out prompting in such a way that they can really get some quality output. But for the most part, what we see, we recognize it.

Jonathan (06:17.082)
Mm -hmm.

Erin Austin (06:36.542)
when we see it, we recognize it when we see it on LinkedIn, we recognize it when we see it in newsletters. We recognize it when we see it on websites. It's right now. It is not something that is reputation enhancing. And so I feel that there's very little ethical breach. You know, I mean, there's, you know, there is a certain level of dishonesty

Jonathan (06:52.506)
Right.

Erin Austin (07:06.046)
I think that if you are using it for something that is truly, you're truly presenting it as being something unique and it's not yours, I think it's dangerous. But I think that from an ethical perspective, that the plagiarism label

shouldn't apply to that because it is AI generated.

Jonathan (07:39.258)
Yeah, that's interesting. I mean, it does make sense. And actually, here's something I want to call out. When I first saw that post, I thought that the original poster, who if I can find that I'll link to in the show notes, but the original poster, I assumed what they meant was that the person using the AI was plagiarizing the people that the AI was trained on.

But that's a different thing. And that isn't what they were actually, that was my first impression. But when I really read it carefully, they were literally talking about crediting, like chat GPT as the author of this text. And, and if you didn't do that, then you were plagiarizing chat GPT almost like an individual. But I, I, I agree with, I think there's a lot of other stuff that we can talk about here, but I agree that crediting, like feeling the need ethically to say,

you know, I credit chat GPT with this feels really weird to me because it's not an individual. There's this other gray area that's right next to this, which is things like Grammarly that is correcting your things like punctuation, but it's also making suggestions about how to how to make a point more clearly and or just a human editor, a human editor.

Erin Austin (08:57.47)
you

Jonathan (09:07.674)
can have a massive influence on, let's say, a book. But then I credit it as the co -author. It's totally different.

Erin Austin (09:11.422)
Yeah.

Erin Austin (09:16.51)
Yeah, well, there are shades of gray here where we're getting into copyright issues, certainly when we're talking about a human editor. So from the copyright perspective, if we want to have copyright protection in something that is AI generated, you probably know, your audience probably knows that AI generated content is not eligible for copyright protection because US copyright

law requires that the original content be created by a human. And AI is not considered human, therefore AI generated content is not eligible for copyright protection. But there is kind of this grading of how much AI input can be involved and still get protection. And so if it's 100 % AI generated, none. If AI is your co -author,

Jonathan (09:57.85)
Mm -hmm.

Erin Austin (10:16.478)
then you can, it does not defeat the ability to get copyright protection in the other parts of it. So say you have a book and you use AI to create some scenarios for you. You want to like, so you can have some examples in the book and you use AI to do that. So you can have your copyright protection apply to all the texts that you wrote, but then you would carve out those parts that were

generated, AI generated, and you would disclose, you know, in your copyright registration application that the, all of the scenarios are AI generated and I don't claim copyright protection for those. Then there's AI as a tool. So if we talk about Grammarly or we talk about Hemingway, I think Hemingway is still a thing, or even like Photoshop. Like if we're using it as a tool to refine our human creativity,

Jonathan (11:04.538)
Yep, that's another one.

Erin Austin (11:14.878)
then that does not, we can still claim copyright protection for that, the entire work. Obviously, you know, how much is too much? I mean, like, there's no, you know, there's no formula for that. But that would be, that's generally the U .S. copyright office's advice when we are registering our copyrights and what we need to declare as being

Original human created versus AI created.

Jonathan (11:47.546)
How much parallel or precedent is there from things like editors or will be another thing? Trying to think of or like, like, let's say you see a lot of sort of business books in the.

Hehehe

Jonathan (12:06.554)
the beginning of each chapter, there's like a pithy quote from Steve Jobs or Sun Tzu or something like that. Are you saying that in a case like that, those those little pieces would be carved out of the copyright of the overall book?

Erin Austin (12:21.118)
that's a good question. Generally, you don't carve out every little quote that is in there. So it would be kind of real chunks of something. So overall, a work, let's just take a book that quotes people. I have never seen someone carve out the people that are quoted inside within the book or quotes that we pull out.

Jonathan (12:48.474)
Hmm.

Erin Austin (12:50.366)
Now, would you be able to get protection with respect to if somebody else took the quote that you used and used it for something else? Obviously, you would not have copyright protection or remedies against that use wherever you got the quote from may or may not have remedies, but you wouldn't. But so it's not, you know, copyright registration is not intended to, you know, that finally parse everything that's in a work. But but

you know, originally, you know, these issues with how much you can use from, you know, from technology and whether or not it can still get registration again is, you know, how much did it assist the human creativity? And I was looking at a, I'm trying to remember why I even looked at this, but it was, it was an early,

issue regarding the use of technology when creating copyrighted works and it involved photography. And so, and this was actually, you know, involved a photograph of Oscar Wilde who happens to be a favorite of mine. And so, so this would have been, you know, kind of 1910 or so early photography and photography. And, and they had taken a photograph of Oscar Wilde and a

Jonathan (14:09.822)
You

Erin Austin (14:21.086)
lithographer just took the photograph and was selling copies of it. And the photographer claimed copyright protection in the photograph. And the lithographer said, no, this is just mechanical. The only person who has any rights in this is Oscar Wilde. The photographer doesn't have any rights in this. They just set up a thing. And that was when, that's the earliest that I have looked at where the US copyright law

Jonathan (14:40.89)
Hahaha.

Erin Austin (14:49.438)
applies to using technology to create something that is eligible for copyright protection. And so that kind of flows down to where we are today. But yeah.

Jonathan (15:00.666)
So where did that go? Was the lithographer found to be breaking?

Erin Austin (15:06.494)
Yes. So photographers own the copyright interest in their compositions. So the idea is that, you know, the way you pose someone, the lighting, you know, the exposure, those things are all creative. And that's the creative work of the photographer. And the photographer does have copyright ownership of the photograph. You know, if there is a person in it, that's a separate issue.

whether or not you have consent to use the image of the human in there. But yeah, but the photograph, the composition itself is eligible for copyright protection.

Jonathan (15:45.914)
No pun intended. So I keep on wanting to find parallels to pre -AI to, I guess I should reveal my hidden biases. I'm a pretty big fan of generative AI at the moment. And I see a lot of fear mongering from people who are.

Erin Austin (15:47.422)
Hahaha!

Jonathan (16:11.13)
you know, afraid of getting disrupted or whatever the case may be, but it really feels like major progress to me. I really think it's, you know, I could end up regretting these words, but I feel like unbalance is going to be a good thing. I don't think Terminator is going to come and kill us all. So, so I'm, I'm looking for parallels where this is where stuff like this is already fine. That was, that were just pre I pre AI. So, you know,

Dear listener, take that. So take all this with a grain of salt because I've got a horse in the race. But what's the copyright situation with someone who you hire someone to ghost write a book for you?

Erin Austin (16:52.254)
Yeah. So with copy, with ghostwriting, you know, again, the U .S. copyright law, the default without anything in writing is that the human being who created the work is the copyright owner. And so in the instance of your ghostwriter, if there's no written agreement between you and your ghostwriter, your ghostwriter owns what he creates and you have a license to use it.

But so if we want to change the default, it has to be in writing and it has to be signed. It can't be by email. It can't be oral. You know, you can give a license orally, but you can't change ownership of copyrights without something in writing and signed. And so if you want to go back to pre -AI concerns about using lots of

Jonathan (17:38.074)
Mm.

Erin Austin (17:47.262)
information and how the rights flow. OpenAI and probably some of the other early gen AI platforms relied on an old case regarding, is it Google Books? Where they went through and basically copied libraries and libraries of books so that they could be searchable via Google. And so

It was originally considered, was thought to be a copyright infringement because you are copying somebody else's work without their permission. But it was held to be fair use. And I hate even bringing up fair use because it's such a loaded topic. But basically the courts determined that, you know, scanning all of this information into Google's database so that it is searchable is transformative.

Jonathan (18:28.826)
Mm -hmm.

Erin Austin (18:43.646)
and that Google isn't actually selling you the book, it's just helping you find the book and then you can go off and get the book from wherever legal sources you get it from. And so OpenAI and its colleagues kind of use that as the reason why they could basically scrape the internet of all of its data in order to create its trading data sets. That is to be determined. That is in courts right now.

And so, but I will say that, you know, going forward, they aren't doing that. They're, you know, getting licenses from people who have massive amounts of data, you know, like, you know, newsletter archives, newspaper archives, you know, yeah, film companies, you know, people with radio archives, you know, that kind of stuff and getting licenses. But yeah.

Jonathan (19:28.762)
Reddit.

Jonathan (19:35.258)
Yeah. Yeah. There's like a new lawsuit every day. So, but that's, but see, that's the, that's not what we're talking about. So the, that fight between the training, the training data people, the people whose data was used for training is, that's definitely a different thing. Right.

Erin Austin (19:51.934)
Yeah, there's two sides to, you know, AI. There's the input, which is the scraping the internet or getting a license from, you know, a newsletter, a newspaper, and there's the output. And so the output is we have the issues of copyright infringement, you know, because the input, we don't know where it came from because, you know, how can you know where a billion data points came from?

But we do know that it can have outputs with the right prompting. That is essentially a copyright infringement. And so we have risk when we are using the output without knowing what the source is and not kind of adding our own layer of creativity to it to make sure that

if in fact this is something straight from somebody's archives that we aren't committing copyright infringement.

Jonathan (20:58.266)
And that piece though is totally independent of how it was generated, right? Like in other words, if I typed in an elaborate prompt and it outputs the exact text of permission marketing by Seth Godin and I published that as a book with my own name, it doesn't matter that an AI did it, right? The real problem is that it's the same thing.

Erin Austin (21:23.102)
That is right. It is you are, if you publish it, you have committed copyright infringement. And by the way, even if you credit Seth Godin, it's still copyright infringement because one, you know, the plagiarism, you haven't plagiarized because you've given him credit, but you have used his protected work without his permission. So when I am the owner of a copyright, I have the exclusive right to

distribute it, publish it, make copies of it, make derivatives of it, grant other people rights to do those things. And so if I take something that was spit out of a chat GBT and it happens to be directly from somebody else's work, I know that chat GBT doesn't have the rights to use it. And they don't pretend to, if you look at their terms of use, they don't claim to grant you any rights in the output.

And then I use it. I have committed the copyright infringement because I do not have permission to use his work. Now, you know, without getting into fair use, because there's always that issue. But, you know, but if we were to take that and put it in our course, you know, and we're selling our course with somebody else's work in there.

copyright infringement and it's not a matter of intent. It is a, you know, did you copy somebody else's work and did you have permission period? And if you didn't and you copied then as infringement.

Jonathan (23:01.914)
Yeah, so it's it would it should be scary for both the plagiarism and the copyright reasons to just use something verbatim that comes out of one of these things. So let me jump back a little bit to the ghostwriter thing where you said that, you know, if you had something, you know, if they had entered into a contract, so like the ghostwriter and the person who wants to pretend to be the author of a work. I mean, Chachapi T.

Erin Austin (23:14.558)
Absolutely.

Jonathan (23:29.85)
An AI can't enter into a contract, but would you... There's no one entering a contract with... It wouldn't even make sense to me to have a contract with OpenAI for... It doesn't even make sense.

Erin Austin (23:45.406)
Right. So that's where we get to the who is it copyrightable? Is it eligible for copyright protection? So is is the is AI the co -author? And so in that case, I don't have to give the credit or enter an agreement with it, but I just need to not pretend to have a copyrightable interest in the part that AI generated. So, you know,

I don't think you have an ethical obligation to disclose that some part of it was generated by AI, but from a legal perspective, you will not have copyright protection from the parts that were AI generated.

Jonathan (24:30.698)
Okay, so I want to jump back to another thing that you mentioned early on. We're probably I don't know for around the same age, but I definitely grew up with the Encyclopedia Britannica as my main source of that was my internet back then. And I don't recall them referencing anyone in those articles. I mean, maybe maybe they had footnotes, but I don't think so.

Erin Austin (24:40.638)
Yes.

Yeah

Erin Austin (24:51.838)
You're right. I don't think they did. I mean, I believe they did original research, but I'd have to look to see if they did. But I believe, now I am talking by the way about the paper copies of it, the hardback copies of it. Okay. And I believe they had researchers who compiled those. But I know that definitely predates my interest in this topic.

Jonathan (25:06.138)
Yeah. Yeah, yeah.

Erin Austin (25:19.454)
But I can look at that. I can look at that to see how they did that.

Jonathan (25:19.578)
Yeah, it was not crawling with footnotes. So, okay, so what do you...

Erin Austin (25:24.638)
No.

Erin Austin (25:31.582)
And by the way, let me just, because not everything we write is eligible for copyright protection. So facts, like world history is not eligible for protection. Now the way we express it, the way we write about it is, but all the facts of history are not protectable. So trying to think back to how much of it was creative and how much of it was just information would also depend on

Jonathan (25:37.498)
Mm -hmm.

Erin Austin (26:02.174)
whether or not there were copyright issues there.

Jonathan (26:05.146)
Right. Yeah, it's I mean, I get my stuff gets ripped off all the time. Like like so. And it's sometimes more blatant than others. And. Of course, I've read a million books about business subjects. I write about business subjects and. But at a certain point. Like here's a specific example, I remember.

probably 25 years ago, maybe more, I read the E -Myth Revisited by Michael Gerber. And about 15 years later, one of my cousins was about to start a new business and I recommended that she read it. I was like, I remember this book was great. It really helped me at the time.

and she was in the same kind of phase that I was in when I had read it. So I recommended it to her. But before I recommended it, before I decided to recommend it, I wanted to reread it to see if it held up because it had been so long. And when I reread it, I was like, my God, there are so many ideas that I have that came directly from this book and I didn't even remember. You know, not verbatim, but like, wow.

Erin Austin (27:16.958)
Mm -hmm.

Jonathan (27:22.234)
I strongly believe this thing, it seems obvious to me now, but I didn't remember that I learned it from that book. And I'm not talking about plagiarizing or anything like that, but at a certain point, you like learn stuff from other people and it gets processed and then it's gonna almost like an AI. I feel like there's a similar argument here where I've read at least a thousand books and the ones that were good had an impact on me. And then that sort of like

Erin Austin (27:34.59)
Absolutely.

Jonathan (27:50.362)
jumble of a thousand, you know, or maybe 10 ,000 ideas comes out in a Jonathan Stark version. And I think it would, I don't know, it would be not just problematic, but impossible for me to even trace all that back. So yeah, at what point does the idea become, you know, go through the food processor enough for it to be a new idea? Cause I, cause writing about sales and marketing and things like that.

Erin Austin (28:06.334)
Yeah. Yeah, well... well...

Erin Austin (28:16.894)
you

Jonathan (28:19.962)
There's no new ideas.

Erin Austin (28:21.246)
Right, absolutely. Well, first of all, ideas are not protectable. And so, I mean, you can protect them if you keep them confidential using a confidentiality agreement, but anything we're reading, those ideas are free for us to use. What we protect with copyright law is the expression of the idea. And so the more, I don't want to say generic an idea is, the less protectable the expression will even be.

So let's say if we're talking about roses by any other name, that is a very unique way of expressing something. And so we can talk about the idea that no matter what we call something, the name doesn't matter. But if we say it in such a way that it evokes a very creative way of talking about that, then that is different than taking an idea like all leaders

are good listeners, you know, and, and so that idea, we can all use it. If you write a book about someone talking about great leaders being good listeners, and they wrote about it and kind of plain business English, and you write about it and playing, but you know, there's, there's, there's, it's going to be, it's unlikely becomes less likely that you will have copied their expression

in a way that is so similar without being generic that would give rise to a copyright infringement. So it really kind of depends on the nature of the underlying work. Like how creative is the underlying work? So stealing the idea of a boy wizard with a, you know, lightning bolt and living under the stairs, you know, you can change all the names and

Jonathan (30:08.506)
Hmm.

Erin Austin (30:19.39)
places and the idea like okay you've stolen the idea there but if I'm

Jonathan (30:23.962)
But you're saying that's maybe plagiarism but not copyright infringement.

Erin Austin (30:28.478)
Well, there's an interesting, under copyright law, you can have a character copyright. And I probably shouldn't brought that up. I was like, I'm gonna get myself in trouble here. But you can, so there you can have like a character that's so well known that you can have protected interest in a character, but that's pretty rare. Those are the Harry Potter's of the world kind of thing, you know, so yeah.

Jonathan (30:50.202)
Okay. Right, right.

This fascinating. It seems like it does. Now all of this. Well, it turns into a book. So I was going to say, like, is the fact that a lot of I guess probably the vast majority of copy I get copyright is the right to copy, not copywriting. So how much of an effect does the digital nature of all of this stuff come into play?

You know, like with I guess either plagiarism or copy or copy or either way, I guess copyright is more important one. But, you know, like like today, somebody sent me a picture of one of my comics that someone just completely ripped off and put their name on and a little copyright symbol with their name and backdated it to before I was even doing comics. But.

Erin Austin (31:41.022)
Mm -hmm. Mm -hmm.

Erin Austin (31:50.558)
Wow.

Jonathan (31:52.154)
I know I didn't rip his off. So it seems like kind of a big coincidence.

Erin Austin (31:53.214)
Alright.

Yeah, well, I mean the big difference is the just the ease right? I mean, it's so easy to do it You know, we always had access to anything that anyone published, you know If it's published and therefore made public we have access to it and we can steal it But there's a lot we need to do if it's you know we took it out of the Sunday paper and we've got to retrace it and do all these things versus if we can download it and

Jonathan (32:03.354)
Mm -hmm.

Erin Austin (32:25.118)
I assume there's some sort of AI, something that can change it for them. And it is the ease, but the fact of copyright infringement hasn't changed based on the technology. That if they are taking your work and they're creating a derivative of your work using technology or using their hand, that is an infringement of your rights as the copyright owner of that work.

Jonathan (32:54.298)
Yeah, I mean, honestly, I think they probably just do more damage to themselves than certainly than to me. But the only thing that makes me nervous about it and and I don't want to pull us into an off topic thing, but I suppose it's closely related is I've I've heard that you can kind of be copyrighted out of your own stuff. Where I don't know if if like, say this person who's who's

is stealing my content, or maybe it's just the AI company hoovering it up. If the original creator didn't do anything to kind of say, yeah, this is mine. I created it on this. I created this thing on this date. And then someone quickly rips it off. But and you didn't do that. And then someone quickly rips it off and then does that. Then you can't like, for example, sell your own book anymore because some are your course. Like somebody takes a course and and and

Erin Austin (33:38.846)
Mm -hmm.

Jonathan (33:53.082)
generates a version of it with like their face in the videos instead of mine and their voice instead of mine. And I didn't do anything with the I don't even know who I would go to with the copyright department, the copyright authorities, you know, like, and then and then the other person sues me because they say that they're the ones that have the copyright. That's the that's I mean, normally, I just I share things with like MIT very permissive license.

Erin Austin (33:57.758)
Mm -hmm.

Erin Austin (34:07.55)
Okay. Exactly.

Jonathan (34:23.098)
so that they do spread. I want people to share them and I don't, you know, it's nice when you get credited but I don't care that much. It's the stuff I'm publishing for free anyway and I want the ideas to spread. That's more important to me and I'll make my money in other places. But those other places where I make my money, someone in theory, I suppose, could beat me to the punch and block me from selling my own stuff or is that just totally

Erin Austin (34:45.566)
Yeah, well, I do encourage copyright registration of your moneymakers. I mean, like, you know, I know you actually write every single day and publish every single day and you are not going to register everything that you write. But when you do have those moneymakers, I do encourage you to register in the copyright office because that provides prima facie evidence of your

ownership of that work. And so anything that was created after that, you know, regardless of when they date it, would, you would have the ability to enforce your exclusive rights against, you know, those infringers. You know, copyright protection does attach automatically the moment you create it. But if you don't register it in the U .S. Copyright Office, then you do have

some exposure for somebody else creating something that they claim is their original and registering it ahead of you. That could happen. But you know, again, like, nah, you're not gonna do that with everything. But with our money makers, then I would do that.

Jonathan (36:01.338)
Back when I was in music school in the early 90s, late 80s, early 90s, all of us songwriters used to make cassette tapes of our songs and mail them to ourselves. Yeah, yeah. And that actually brings up music, which is another one that I don't know if you were. We've been talking mostly about words and a little bit about photography. The music one, I feel like, is another one where there's a lot of.

Erin Austin (36:10.334)
You know, they called it the poor man's copyright, I heard. Yeah, yeah.

Jonathan (36:26.938)
what seems like it could be precedent for remixing things and you know like DJs and sampling and all of that stuff.

Erin Austin (36:37.854)
Well, you know, that brings up kind of a parallel. You know, when we look at how to fix AI, you know, when there's billions of data points out there, how are you going to compensate all the people who may have, you know, rights in the data sets? And the some of the solutions are looking for is like a compulsory license, which is what we have, you know, in music. So that if I want to, you know, use one of your songs and it's with the

with a music publishing house and I can do that. I don't need to come and get your permission to record it or perform it. But.

Jonathan (37:20.025)
Well, what's so like, is that like a cover band can play Stairway to Heaven in a bar? Or do you mean or you don't? actually, you could record a cover. And I think the copyright is for the recording, not the song or I don't.

Erin Austin (37:32.638)
Well, there's two. We are getting a little bit dangerous, but we have a couple of levels of copyright. So you have the song. So that is the, the lyrics and the notes, right? And so whoever wrote those, they are, that is protected under us copyright law separately. We go into the studio and we record it. And the recording is a separate work that has,

separate copyright protection and may even have a separate owner, like such as a recording studio, maybe owns it. But what permissions you may need to exploit that recording will depend on, you know, who wrote the music, you know, who wrote the lyrics and the music and and who performed it.

Jonathan (38:07.194)
Mm -hmm.

Jonathan (38:27.706)
So here's like, I wonder if this would, if anyone's considering, I'm sure someone is, but the, you know, ASCAP and BMI, I assume they still exist. I have not really paid attention to this in many years. But if you go into a, dear listener, if you go into a bar or something and you know, the radio is on or jukebox is playing stuff, that...

Erin Austin (38:40.51)
Yeah, I believe they do.

Jonathan (38:53.37)
that establishment is supposed to be paying some amount of money to this trade association that then distributes the money in all these crazy ways that Erin just described. There's like songwriters and there's publishers and there's music labels and there's recording studios and there's a million things. And they somehow trying to find that money up across all those people. I know it certainly has been a big problem over the years.

But that feels kind of similar to what's going to is that kind of what's happening on the other side on the on the open AI fighting with the original creators side of things where it's just going to turn into like, well, we'll just kind of like split it up and try and try and spread it out.

Erin Austin (39:25.886)
Yeah.

Erin Austin (39:38.75)
I mean, that has been suggested because at the end of the day, you know, AI is too big to fail. Like, I can't scrape all back the billions of pieces of data, you know, and figure out where they all came from and individually license them. And, you know, it's just it's impossible. And so, you know, what

Jonathan (39:57.722)
the horse is out of the barn.

Erin Austin (40:07.262)
can be done to fairly compensate the content that they got for free. And so yes, that's something that's been suggested. I don't think we're anywhere close to figuring it out. And that only the big data owners are gonna profit from this. But you know, you know.

Jonathan (40:28.634)
Right. And yeah, and you look at a platform like Reddit where the owners of Reddit didn't create the user generated content. Like the the

Erin Austin (40:33.342)
Yeah.

Erin Austin (40:37.918)
Well, meta, I mean, that's the big thing right now, right, that they are now taking your photos if you don't do all the right stuff. It's going into data sets. So.

Jonathan (40:47.213)
Yeah, so, so well, I think, I mean, if you look at Reddit, which isn't one of the companies building an AI, right? Well, who knows? Well, they probably aren't because if they were, they probably wouldn't have sold their data. So, but they're selling, they're not selling their data. They're selling user data or user created data that.

Erin Austin (40:56.862)
Mm -hmm.

Erin Austin (41:07.55)
That's right.

Jonathan (41:12.09)
In theory, like the Reddit users should be the ones getting compensated enough, but I'm sure in their terms of use or whatever, it's like everything you type in here is ours.

Erin Austin (41:20.574)
Absolutely.

Jonathan (41:22.394)
So, yeah, so you're right.

Erin Austin (41:25.694)
Yeah, I mean, you need to, if these, these are your concerns, you need to read your terms of use. I mean, no one ever, myself included, reads the terms of use generally of the platforms that I, the social platforms that I use. I've never read the terms of use of Facebook or Instagram or LinkedIn for that matter. and, and you need to read those things if you want to.

figure out how you can prevent your content from going into being sold or licensed for AI use.

Jonathan (42:01.082)
Yeah. And, and hilariously, I don't know who said it, but if it's free, the product is you. So, you know, I'm sure if you looked every single one of these platforms is like, gold rush. And, you know, we, the selling data to people is nothing new that's been happening. Certainly probably before the internet where people would collect, you know, mailing lists of people's just physical addresses.

Erin Austin (42:06.526)
That's right.

Jonathan (42:28.538)
So I'm sure every Terms of Use you go to on any one of these platforms is going to be pretty hostile to the users.

Erin Austin (42:35.262)
Yeah, I mean, we know that they've been taking our contact information and selling it to advertisers where we get followed around the internet with something that we looked at for a second one time on our phone. But now, yeah, they're actually going in and grabbing our family photos and things now. So, yeah, yeah.

Jonathan (42:56.378)
I had not heard that one yet, so I'm sure that's... That seems like that would cause a really big outcry from the user base because that seems even... Honestly, it's a drag when someone rips off a comic strip or something like that, but like baby pictures? That's...

Erin Austin (43:12.542)
Mm -hmm.

Yeah, well, I mean, here's the thing about AI, though. I mean, they are collecting these billions of data points, right? And they're using it to generate something new. That is their mission. It's not to retrieve your baby photo. It's to use it as a data point so that if somebody asks me to create a photo of a toe -haired boy, then they'll go and say, OK, here we go, you know.

And so it should not be that your baby photo is popping out. But yeah, but that, and that same with people who are really, really worried about them scraping something off of your website. Like, yes, they can, but members going into a data set among billions of other pieces of information. And so if somebody really wanted to rip off your website, well, why don't they just go to your website? Yeah, there's a bunch of easier ways to do it.

Jonathan (44:00.874)
Hehehehe

Jonathan (44:11.93)
easier ways to do it. Yeah. Right. Yeah, the baby pictures being private. And I mean, basically, what we're talking about is like the biggest law machine of all time is taking all this stuff and chopping it up and into bits. They're so small that nothing super recognizable should should be able to pop out the other end.

Erin Austin (44:15.678)
Yeah.

Erin Austin (44:35.742)
Right. Right.

Jonathan (44:41.498)
But yeah, I don't know. Where do you?

Erin Austin (44:43.198)
And there are some, especially with respect to images, AI that generates images, they're starting to put in protections to make sure that nothing literal comes out. I don't know how the technology works, but that is very important, because it's important to everyone who's creating images using AI. So I think that, at least from that perspective, you're not going to get someone's

baby photo popping out.

Jonathan (45:13.658)
Right. Yeah, there's some stuff in there. I typed, actually, my daughter typed in something like, we're she typed in, give me a Gen Z version of the Declaration of Independence. And, and it was like, yo fam. It was it was hilarious. We

Erin Austin (45:32.862)
Erin Austin (45:39.23)
Yeah.

Jonathan (45:40.09)
We ran it a few different times with different generations. It was super funny. It was very funny. And then she wanted a picture of Obama dressed like a millennial or something like that. We went, it was like we were having a political conversation already and it was like, nope, we wouldn't generate any kind of picture of Obama. And I'm sure there's some.

Erin Austin (45:44.094)
That's... that is funny.

Erin Austin (45:52.894)
Yeah

Erin Austin (45:59.646)
That's right. Yes.

Jonathan (46:03.93)
There's some layer there that's very, very slapped on that is filtering for celebrity names and celebrity brands. Like I'll bet you if you tried to get something that is, is under the Disney domain, you wouldn't get it. You know, like Mickey Mouse or something to be like, yeah, no, you're not getting that.

Erin Austin (46:10.942)
There is.

Erin Austin (46:18.238)
Absolutely.

Erin Austin (46:22.014)
Yeah, I saw an example recently. It was Mario Brothers. And they had this kind of very elongated, long, narrow, lanky type of image that came out. But still, same colors and mustache. But yeah.

Jonathan (46:25.69)
Mm -hmm.

Jonathan (46:43.017)
Yeah. I remember using mid journey not too long ago and it was I asked for like a graphic in the style of some artists.

I can't, let's just, I don't know. I can't even think of an artist like in the style of Van Gogh or something. Like, you know, give me a picture of a guy going up an escalator in the style of Vincent Van Gogh. And it's like, we're sorry. We can't, we can't copy styles like that. And then you see, just say, well, like describe the style that Vincent Van Gogh painted in. And it gives you this long paragraph. And then you say, okay, create a picture of a guy going up the stairs with the following attributes. Totally does it. Yeah.

Erin Austin (46:59.358)
Mm -hmm.

Erin Austin (47:19.39)
And it does it okay interesting

Jonathan (47:22.554)
So it's not very, it doesn't seem very sophisticated yet, but, you know, hopefully, I mean, hopefully we'll get to a place where like the fear mongering, which I'm not saying it's baseless, but the level of, of just sort of, you know, FUD that is in the media and the headlines, it's like so easy to point to this monster.

Erin Austin (47:36.19)
Mm -hmm.

Jonathan (47:52.442)
And I, you know, I just have a lot of people, family type people just like, just have this vague fear about AI. They don't use it. They've never used it. They don't understand what it is. And even when they do, if like, well, here's chat, TPT on my phone or whatever, and they don't do it. They like don't know what to ask it. And they kind of Google with it. It doesn't, and it doesn't like, what's the big deal? And there's just so, it's just, it's so classic.

Erin Austin (48:00.734)
Mm -hmm.

Jonathan (48:22.874)
of this kind of radical new technology and the again, there's definitely lots of problems and but I just think it's going to get worked out. You know.

Erin Austin (48:25.854)
Mm -hmm.

Erin Austin (48:34.11)
Yeah. Yeah. I think so. I mean, obviously it's effects every, I mean, whether we're aware of it or not. I mean, they have used it. They think they haven't used it. They have used it because it's everywhere. Right. And it has been for some time. We just didn't realize it.

Jonathan (48:43.866)
Yeah, they have, right. Yeah, if you've taken a picture on a new phone, yeah, so it's been corrected. You don't look that good.

Erin Austin (48:55.966)
Yeah

Jonathan (48:59.546)
Filters are on everything now. So yeah, I mean how? Yeah.

Erin Austin (49:03.102)
Yeah. I mean, it all feels very personal to you, but it really isn't. I mean, it's not any more personal than, you know, Anson and Anhill. I mean, it's just, it's, it's just, we are just a little speck. Our information is just a little speck in there. So.

Jonathan (49:22.49)
Yeah, it's hard to picture trillions. So it's just no perspective on it. I'll probably end up regarding all of these words, but I've always been kind of optimistic about new tech, perhaps overly optimistic. Everyone was freaking out about mobile. Everybody freaked out about, you know, you name it, comic books. Socrates was against literacy. I mean, it's like...

Erin Austin (49:26.11)
Right.

Erin Austin (49:42.238)
Mm.

Jonathan (49:49.69)
Yeah, it's like, the people are going to it's going to ruin people's memories if we just write stuff down. And maybe he was right. But, you know, I only know he said that because someone wrote it down. So, you know, like the irony.

Erin Austin (50:00.254)
Right. Well, think about all the good things. I mean, all the things like access to lawyers, you know, that only a certain segment of the population can afford. And you have, you know, AI assistance or even, you know, healthcare to have AI, you know, helping with imaging and interpreting, you know, and diagnosing things. I mean, there are some major advancements that I think will be very

good for mankind. But yeah, it is the Terminator scenario that we worry about, right?

Jonathan (50:40.538)
Yeah. Yeah, I think the near term one, the one that I think is most founded is the job disruption. Like that's definitely going to happen. And it's kind of like, you know, how, how, you know, the plane is going to land one way or the other, but you just don't want it to go nose first. You want it to have a nice smooth landing. So, so I, my prediction, if people are wondering, my prediction is that the job thing

Erin Austin (50:50.238)
Mm -hmm.

Erin Austin (50:58.59)
Right.

Jonathan (51:10.106)
is going to be a, it's not going to be a crash. It's going to be a slow transition that, that fingers crossed takes long enough that people who are growing, you know, like in high school now with AI tools will just go into a different direction than they would have otherwise. And it's not like people will be getting fired because, you know, well, now we have chat GPT, we don't need you.

Erin Austin (51:22.014)
Mm -hmm.

Erin Austin (51:33.982)
Right, we don't need court stenographers anymore or whatever. Yeah, my son just graduated from high school last week and the principal, yeah, was talking about some of the areas of study that some of the students were going into and he was all listing those that weren't around 10 years ago, you know, and so it is, everything's evolving, right? So the beast needs to be fed.

Jonathan (51:36.474)
Yeah, yeah.

Jonathan (51:41.274)
Congratulations.

Jonathan (51:59.382)
Yeah. Yes, progress is going to happen and that's going to upset some people for sure. And like I said, I could easily live to regret these words, but we shall see. Right now I'm getting a huge amount of utility out of it. And in none of the ways that would violate any of the things we talked about today, there's just so many ways that you can steer way clear.

Erin Austin (52:10.078)
Mm -hmm.

Jonathan (52:25.306)
of things like submitting phony court briefs or just really taking any output and just using it. That would scare me a little bit. I don't do that. I'm more like, it's like, it's so great at taking giant documents that you wrote so you know that it's original and like boiling it down, transforming it into other formats. It's incredible.

It's super useful. It's a great sort of brainstorming partner.

Erin Austin (52:58.526)
Yeah, as brainstorming as, you know, a first draft, I mean, in the legal industry, you know, best practices is to take anything that you get out of it as if it came from, you know, first year lawyer and assume, you know, it's screwed up and you have to put your eye on it. You need to check everything. And, but you still use, we still use junior resources because there's value to using junior resources. That's how we get leverage, right?

Jonathan (53:12.378)
Right.

Erin Austin (53:25.086)
And so we shouldn't ignore this as another form of leverage in our business. We just need to use it responsibly, for sure.

Jonathan (53:33.338)
Well, that was a mic drop. That was probably a good place to leave it right there. Cool. And I do see that we're coming up to the top of the hour. So, Erin, thank you so much for joining me. This has been great.

Erin Austin (53:36.478)
Ha ha

It has been great, thank you.

Jonathan (53:47.226)
Where can people go to find out more about what you're doing online?

Erin Austin (53:50.59)
Yeah, well, I mostly hang out at LinkedIn and I am the OG Erin Austin. I've been there so long. So just, yeah, Erin Austin. And then my website is thinkbeyondip .com.

Jonathan (54:04.634)
Excellent. Alright folks, check it out. You can talk to Erin on LinkedIn. Alright folks, that's it for this week. I'm Jonathan Stark and I hope you join me again next time for Ditching Hourly. Bye.