Ethics of Generative AI

Doug Berger: [00:00:00] Welcome to the latest installment of Brand of Brothers. I'm Doug.

Johnny Diggz: And I'm Johnny. Today we're talking about the ethics of generative ai.

Doug Berger: All right, let's get to it.

Johnny Diggz: And we're back, Doug,

Doug Berger: John.

Johnny Diggz: So we've been building up for a little while about, uh, that we had this, uh, AI sort of ethics framework that we had been working on. I believe we've mentioned it. 'cause our, our follower has been. Been really patiently waiting for this big reveal, and it's, well,

Doug Berger: but we can't just have a reveal.

Right. We kind of need to talk about how we got here.

Johnny Diggz: And you're right, and, and so. Why don't you take us back to where we really first started discussing, um, or [00:01:00] you really, you, you, you kind of expressed to me, uh, that you had some trepidation right over, over whether or not it was ethical. For you specifically to use AI in your capacity as a, as a owner or co-owner of a, of a agency who is doing work on behalf of your clients.

Right. And where, and, and you were, you were grappling, wrestling, if you will, um, internally and discussing with me and, and other colleagues. About where, where this line was, what, what was ethical to do? What did you have to disclose? Like all these things. So tell me where it started.

Doug Berger: So it all started with, uh, the co-founder of the agency that we have remixed.

Um, and so Simon is our chief creative officer and he is. [00:02:00] An early adopter. He's the one who got the agency aligned correctly when it comes to like, web development, for example. And, uh, he was the one who, who spearheaded our, our mobile first, for example. Um, so of course it's no surprise that he would also be on the forefront of.

Creative utilizing tools like generative ai. And so speaking of, you know, calling it a tool, that's where I've ultimately landed. But there was a day in particular, like couple years ago, um, where. Uh, I happened to be in your neck of the woods and I, I reached out and was like, Hey, you wanna meet up for coffee?

And while we were meeting up for coffee, as I'm sure you recall,

Johnny Diggz: it was at the what? Ru Russos. Raso. Raso Raso?

Doug Berger: Yes.

Johnny Diggz: Uh, shout out to Rossos.

Doug Berger: Yeah.

Johnny Diggz: Lake Mary. No. No. Longwood. Longwood. Yeah. 4 [00:03:00] 34.

Doug Berger: Um, really good coffee.

Johnny Diggz: Yeah.

Doug Berger: Uh, so they're not even paying for this.

Johnny Diggz: I know, I know.

Doug Berger: We'll show this to them and be like, Hey.

Johnny Diggz: Look, how many

Doug Berger: can we get a cappuccino? So, um, so yeah, it, it, it, it. While we're having coffee, I'm like, Hey, I am on the fence about something and I need you to knock me off the fence. And so that something was generative ai because like you were mentioning, I I, I had. I have concerns about the use. It's not had, it's not past tense.

It's it's present tense and future tense that I have. Continuing concerns about the idea of utilizing generative AI when it comes to the development and delivery. Of creative that is solicited by a paying client, right? It, but it obviously goes beyond that as a creative, like there are things that I just care [00:04:00] about, uh, ethically, right?

I don't wanna steal someone else's ideas. Uh, not just because, uh, you're talking about IP infringement, but you know, I. I endeavor to be creative. I want to be original.

Johnny Diggz: Right. So you're saying it's, it's there, there's a distinction because there is stuff that, you know, it, it may be illegal, but that doesn't necessarily make it Right.

Doug Berger: Right. Yeah. Um, so really in this realm, uh, it was establishing, you know, uh, am I using. Content that I'm allowed to be using. Right? So where, where did that con, where, where's the source data that is being mined? So that was a really big concern. But the, the, the crux of this, uh, I, I think was, was really with Adobe Firefly, right?

Because that kind of started to, to mitigate that concern about where the content came from. Um. So [00:05:00] I, I think that's kind of the orig, the, the origin story Sure. Of, of, of where we, where we began. And then I think you started asking questions like, well, you know, what, what are the, the important points, uh, to make sure that you're not violating?

Johnny Diggz: Yeah. Like what, what would be, what would be your criteria to say, well, if it's. For example, um, Adobe, uh, uses, uh, their, they, they started using generative AI inside all of their tools. But it started with Firefly, or is that the name of their tool or,

Doug Berger: I'm not sure. So Firefly is when it became. Truly generative where you could use a prompt, right.

However, I would say that we, we had precursors to that when you were looking at, like expanding a background, for example. Mm-hmm. Um, so it, it was able to generate these backgrounds that matched. So for all intents and purposes, it was generative ai. Um, but it was using [00:06:00] the source data. That was right there in front of you to create the extra pieces.

Johnny Diggz: So you like your ex. An example would be you've got, uh, maybe a, a, a relatively tight photo shot of a, a person and you wanted to make it look like that they were in a big field.

Doug Berger: Right, exactly. And so it could basically look at the scene and figure out what it needed to kind of duplicate, but also offset to make it look real.

Johnny Diggz: Gotcha. And so. That, that begs the question. So there's, it seems like there's another layer there too, because if you're using the generative ai, there's, uh, there's a question about ethics as far as, um, disclosure. To your client, do you, you don't typically disclose to them how you're building their website and what tools you use.

You don't disclose that you used Illustrator

Doug Berger: or Right. They, they don't care that we use Adobe [00:07:00] products. Right, right, right. Um, it, they, they might care if we used things like, uh, um. Shoot, I can't even remember the name of the, the software. There's some crappy software that's out there that's become universal.

Um, but any, oh, canvas, Canva. Canva. Yeah. Canva. Canva. Yeah. Yeah. They might, they might care if we're using that crap, but, uh, but you know, Adobe is, is kind of a standard bearer, so, uh, it, it, it would be, it's, it's as expected as a screwdriver is in your, your tool bag. Right. So, um, it, and, and what we're talking about.

Is, okay, well maybe ai, generative AI is akin to a power screwdriver, right? Mm-hmm. Um, except it can do. Way more than that. And so

Johnny Diggz: it's like one of those multi-tool

Doug Berger: things. It it, it, it is a, a Swiss army knife of sorts and, and, and, uh, the creative process. But the question is where does it belong [00:08:00] in the creative process?

And that's something that I was also grappling with. The other component that I was grappling with was. Is this going to replace someone? Do designers just become creative directors? And is that a sufficient replacement? Like when it comes to what we're doing, um, it's re-skilling, right? Do designers just become.

Visual design engineers. And how do they know that they're creating the right thing without the experience that's required of an actual practicing designer?

Johnny Diggz: Well, some, you know, some of the kids these days would say that prompt engineering is creative. Would you agree

Doug Berger: to a limited extent? I mean, as someone who has been a creative director, my job was to ideate and write creative briefs.

Right. Can you use [00:09:00] chat GPT to do that? Absolutely.

Johnny Diggz: Right.

Doug Berger: Is it probably gonna be shit? Yeah. Right. Because it just doesn't have the, the important elements. For example, it's not very good at being able to gauge what is going to, uh, affect emotion, right? It's not very good at doing things like double ante, right?

And so you, you run into all of these limitations that. A skilled technician is still required. So how do you manage that in the creative workflow? So that was another reason I was on the fence going, this isn't a bad thing. This could be really amazing, but. In the design world, you have seen multiple paradigms.

For example, when I was coming of age in as a designer, we transitioned from where we were [00:10:00] doing literal pay st up where it was, uh, stuff that was adhered to a cardboard that would then actually be, uh, photographed. Um. To desktop publishing. Right. And so desktop publishing all of a sudden made our process super fast.

Johnny Diggz: And we're talking about not that long ago, we're talking

Doug Berger: 30, 30 early nineties. Yeah.

Johnny Diggz: Okay. That

Doug Berger: was like the late eighties, early nineties. Okay. And so

Johnny Diggz: Print Shop Pro,

Doug Berger: uh, yeah. It, it, it really, it when, when you have the first Mac mm-hmm. Um, it, that the, that kind of the Mac. Classic. Right. Um, that's really when it, it, it began to take off.

There was already desktop publishing. Right, right. Um, it, it, in, in the early eighties, we had things like Carell draw, but you weren't really doing desktop publishing with that so much as you were doing like vector art that would get cut.

Johnny Diggz: So you, you talked [00:11:00] while we're back in the, in the early eighties, I mean that was a big revolution with.

In art as well. Right. So we had to transition from, um, you know, into the first, you know, digital art.

Doug Berger: Right. We, we definitely began seeing traditional art. It, it, I would say that started certainly in the eighties where we began seeing digital art. Um, it became a little bit more prevalent. And

Johnny Diggz: I bet people didn't think that that was art.

Doug Berger: No. It was definitely questioned because Right. You know, what kind of skill is involved. Yeah. You know, and

Johnny Diggz: probably the same with digital photography. Back when they went from,

Doug Berger: yeah,

Johnny Diggz: and probably from photo, from, from paintings to photographs.

Doug Berger: Right. So, so I would, I would argue that the differentiation is craft, right?

So when it comes to creating artwork, does the medium matter? Does it matter how much the artist crafted that medium? Did they actually have to get substrate [00:12:00] and pigments and mix it together in order to make their paints, or did they buy it off the shelf? And does that change what type of artist and how good of an artist they are by the level of craft that they employ?

Johnny Diggz: Did Picasso make his own paints? Uh,

Doug Berger: probably and probably not, right. Um, because of, of that time,

Johnny Diggz: right?

Doug Berger: He, you, you generally would have to mix your own paint, but regardless, he couldn't

Johnny Diggz: just go down to Sam Flax.

Doug Berger: Again, not an advertiser, but a fantastic art supply store,

Johnny Diggz: colonial Drive. Florida.

Doug Berger: So a, as far as, uh, as far as the, the art is concerned, right?

Like you were saying, digital art was not widely adopted until the turn of the century. And by turn of the century, I mean turn of the millennium into the two thousands, right? And so in, in the two thousands. Uh, we had already seen [00:13:00] deviant art come up as a platform for digital art. Um, was it what you would find in a museum?

No. Mm-hmm. But back in, in, in the, the eighties, there were very few museum quality digital art out there, right? I think, uh, Chuck Close may have done something. I have no idea. Don't quote me on that. Um, but. But it, it, it was people like that. Uh, and, and April Gron, I think was her name. Um, it, so there were, there were some, some digital artists that were noteworthy in that era, but it didn't become mainstream until decades later.

And I think that's kind of where we are right now. We are on the, the, the precipice of generative AI going from being. A, a niche space to something maybe 30 years from now, I would not be surprised to see, uh, displayed in a gilded frame, in a [00:14:00] place like, uh, the Metropolitan Museum of Art, for example.

Johnny Diggz: The, it's interesting to see these sort of inflection points where, you know, I was just thinking about, uh, you know, digital versus, uh, digital, digital photography versus, uh, analog photography or chemical photography.

Um, and you know, the same goes in music with samples and I mean, there was a huge. Uh, controversy back in, in when sample sampling became popular and the legal ramification, it took years and still evolving as, as a law. The law's always trailing behind the technology, um, and seems to be getting further and further behind.

Um, and so it's important that we don't just look at what's legal mm-hmm. Um, that we look at what's ethical and. Um, and from the, the. Ethics discussions. There's a lot of ethics discussions [00:15:00] around the, uh, the, the source of the data of, uh, where they're trained, where these models are being trained. So, you know, uh, New York Times has, uh, filed a suit against open ai.

Um, and, and there's countless other lawsuits going around and the suno is being sued for training on on, mm-hmm. On music and, and, uh, and I'm sure there's many, many more, um, about this, this training data.

Doug Berger: Yeah.

Johnny Diggz: Which, which is sort of, is, is separate from the use. Ethics. And so let's talk a little bit about this, the, the training data as you see it.

Um, we can bo both talk about the, the, on the,

Doug Berger: I mean this is definitely more of your wheelhouse.

Johnny Diggz: Yeah. Well, I was talking on, well, like for example on the legal side.

Doug Berger: Sure.

Johnny Diggz: Um, you can, you have a legal responsibility to your clients. [00:16:00] You, you have an, you, you, they engage in a, an agreement with you. You're performing work on their behalf,

Doug Berger: right?

Johnny Diggz: You, and, and so they, you transfer a license to them. That, that, do you agree? Do you don't transfer the whole, like how does that licensing work today? Not including generative ai, like if you just did a logo design and all that work mm-hmm. For a client, um, the old fashioned way.

Doug Berger: Yeah.

Johnny Diggz: Uh, how, how does that on a legal side, I'm talking,

Doug Berger: so on a legal side, the, an artist always.

Maintains the copyright unless they explicitly sell the copyright.

Johnny Diggz: Okay.

Doug Berger: Um,

Johnny Diggz: that's true of a photographer or, yeah. Yep. Okay.

Doug Berger: And, and so what we do is as a business, and this is a business decision. Yeah. Um, there are many businesses that do it way different from us. Yeah. But when we work with a customer, we believe what we're doing is for them and them alone.

And so [00:17:00] they get an indefinite. License to

Johnny Diggz: Right.

Doug Berger: Utilize whatever it is that we provide to them. In perpetuity. And that is until the death of the company or if they're acquired. And if they're acquired, then that license transfers in that acquisition. Um, so, but that's not the case with every agency or designer.

Some designers have limited, uh, licensing. And the same is true for photographers, right? A photographer might only do a, a one year. Allowance and might also limit the applications of said photo, right? So that photo might be, uh, something that you're only allowed to use on a website and in a print campaign, and, uh, and you know, a digital campaign and you can only use it for a year.

Sometimes a photographer will just be like, here are the negatives. You can have [00:18:00] whatever. Of course, you know, figuratively speaking, no negatives, but, but you get the idea. Um, so as far as the transfer of rights is concerned, uh, it, it, it really varies from situation to situation.

Johnny Diggz: So, but, so adding. If you are not the original artist.

Doug Berger: Yeah.

Johnny Diggz: In the case of, uh, AI generated something, um, that kind of puts that entire structure under. A potential weakness. Right?

Doug Berger: It definitely does. Yeah. Yeah. Which is, but it, it's not to negate the fact that there are times as designers, and this is, this goes back to before the term graphic designer even existed, where the, the commercial artist would use a bank of clip art.

Right. And that clip Art might be a book that they purchased that has a royalty free, or it might be something that's licensed. Uh, it really depends. And they would literally cut it [00:19:00] out. They would clip the art and paste it onto the paste board. Um, and so did they originate that? No, not necessarily. They can use a book of illustrations and there are all sorts of books like that, right?

So that have ding bats and, uh, as, as Simon would call them, doilies, uh, and, and just things that are, are, are beautiful finials and, and whatnot. Um, so where, where is it art, where is it craft? And, and does that actually matter? Right? I don't design all the type faces that I use. I license them. And I licensed them from someone who is a type designer.

And I think that's kind of akin to having an electrician and an apprentice and a whole team that is able to do wiring in a building. Uh, I think that the designer is just one of those, uh. Craftspeople that is involved in, in the [00:20:00] creation and execution. So it, I don't know where I'm going in this meandering thing, but, but really the, the big idea is I don't believe that generative AI is.

It is negating what a designer's contribution is, right? Because they're, I use typefaces, I use stock photography or photography from a photographer that isn't me, right? Um, and so there are all of these elements that come together. So using generative ai. Especially if it's using royalty free or, or licensed photography from which to create that media.

I feel like that's not going anywhere. That is a, a, a negative,

Johnny Diggz: the, you mentioned typefaces, you know, who,

Doug Berger: can we edit that section out?

Johnny Diggz: But yeah. Um, no. Uh. Our fan de de demands that these appear unedited. [00:21:00]

Doug Berger: Hi Calvin.

Johnny Diggz: Um, the, uh, speaking of, I just wanted to say, we, we joke about that we have more than one fan, and so if you, but

Doug Berger: Calvin is

Johnny Diggz: easily our biggest fan.

Doug Berger: Yes.

Johnny Diggz: Um, and so, uh. If you, uh, if you see this and you enjoy this content, just write a comment or something in their YouTube feed or whatever, wherever you're seeing this and,

Doug Berger: and like, and subscribe.

Johnny Diggz: And you can do that too. But it just, we, we'd like to know who you are because, um, we, we'd like to cater this content to, to you guys, so,

Doug Berger: so.

My big question to you, John, is so why are we even talking about the ethics of generative ai? Is, is there something that, that maybe we did that we might want to tell people about?

Johnny Diggz: So I think that's what, what we're getting to the the, so as, as a result of this [00:22:00] conversation, I. Went back to my house and had a conversation with AI about our conversation.

That's right. And out of that conversation I developed, uh, a sort of, uh, first draft of a, um,

Doug Berger: it was a matrix.

Johnny Diggz: It was a matrix, right? Yeah. Yeah. And, um, and I, and I gave that to you, and then we sort of. Created a first draft of a document. Mm-hmm. That sort of helped guide us. It was really an exercise in us saying, I was like, well, I like to use Suno and make songs.

Suno is had some controversy because of did they ethically source the data? A lot of people are saying, no. And, uh, so there's been lawsuits and there's been, you know, they just made a partnership with Warner Music and, and stuff like that. But, [00:23:00] um, but. We, we started talking about things like, well, intent.

And, you know, is, is the, does intent matter? Does um, you know, yes, yes. Does, uh, you know, is it, you know, how is the, the data sourced ethically? Like we started building out what are these criteria? Like I said, Doug, what would it take for you to say yes. And so we sort of broke it down into this, realized that, that it couldn't really be a yes or no,

Doug Berger: right?

Johnny Diggz: It's, it's, it's a, it's a scale. It's a range. And what is, you know, what is in the, in, in, you know, very, very like. You know, you're using it for social good. It's helping people. Um, it's, it's, it's creating jobs,

Doug Berger: right?

Johnny Diggz: Um, you know, it's, uh, it, it's, it's [00:24:00] ethical resource. Ethical resource data. Yeah, exactly. It's all of these things.

How do we, how do we define that? And, and can we, and, and

Doug Berger: categorize

Johnny Diggz: and categorize it and come up with a, I don't wanna say standard, but a, a scoring system that we could just. So, you know, look at it quickly. Granted, it's not, it is never, never gonna be perfect. We, we decided very quickly, this is ano, this is going to evolve.

And so we're at what? 2.2

Doug Berger: 0.5?

Johnny Diggz: 2.5 now. Um, and, uh,

Doug Berger: but it's finally ready.

Johnny Diggz: It's finally ready. So, um,

Doug Berger: so we created this rubric and then we translated this rubric into a functional utility. Yes. That anytime we have any questions. About the, the ethical application of generative ai, we can go in and use five different dropdowns that answer five different categories of where [00:25:00] the range is for ethical behavior.

Obviously, it's never gonna be foolproof. More importantly. N nefarious actors can always exist. They don't give a shit about it being ethical. Yeah. We kind of talked, they're not

Johnny Diggz: gonna use

Doug Berger: this tool.

Johnny Diggz: We talked, we talked a little bit about, um, about that. It's like, well, if. If somebody is, is, uh, is going to, I mean, we know that there are people out there already using it to create spam bots and, and steal your data and, and you know, there's, uh, create, you know, fake web pages and all, you know, there's lots of nefarious but

isn't.

Doug Berger: But this isn't about morality, this is about ethics. If you, if you have the morality to be a positive actor, then this ethics scale is for you. Is for you. Exactly.

Johnny Diggz: I think that's a good place to pause.

Doug Berger: I think that's Fanta, you know what, we should talk about that on our next episode,

Johnny Diggz: and I think you, this is the first of a series that we are going to be talking about the ethics of ai, [00:26:00] specifically the scale and the different factors.

Uh, that we believe are, are important to look at when, uh, when looking at AI ethics. So, uh, so like, and subscribe and you can, uh, you can see part two

Doug Berger: Thank you for tuning in to Brand of Brothers. Big thank you to our presenting sponsor, Remixed, the branding agency, along with production assistance from Johnny Diggz, Simon Jacobsohn, and me, Doug Berger. We can't forget music by PRO. Speaking of not forgetting, remember to do that like and subscribe thing and find us at BrandShowLive.

com and follow us on the socials at BrandShowLive.