April 19, 2023
Don: Welcome back to the Design and Prosper podcast. This is episode 75 and today we are continuing on our conversation on AI, artificial intelligence. Today we are taking a deep dive into the dark side. Let’s go.
[Intro music]
Kris: Hey, welcome back everybody. So last week we had our first in this series, so this is a two part series. We didn’t realize we were going to have two parts, but we just talked and talked and talked about it. And we realized, yeah, we need to, we need to dedicate a whole episode to the other side of artificial intelligence.
So we talk about a lot of the positive aspects about it, what it can do to help elevate your design business. The possibilities are endless.
Don: Yeah absolutely. We got very, very excited.
But the flip side of that is that there is a absolutely enormous, vast area of unknown. Huge, huge, huge. And when we have such a massive area of the unknown, that can be incredibly scary and a lot of people are intimidated by the lack of understanding. When we understand something, we feel more comfortable.
And there are a lot of us on this planet that don’t even come close to understanding the power of AI, and therefore there’s a lot of discomfort going on.
Kris: Well, most people don’t like uncertainty and this is uncertainty.
Don: Yeah. And the interesting thing is that in our last episode, we talked about Sam Altman, the CEO of OpenAI, and we have said, if you don’t know his name, you will, because he’s, he’s an incredible person.
And we will link a recent interview with Sam to our show notes for this week as well. We also linked it to the last episode as well because that interview sort of gave some really clear insights into how this is a big big beast I’ll and the makers are just working out how to harness this thing. Mm-hmm. So they’ve let it go out into the world and it’s in beta and it’s being tested and it’s growing at a rapid rate of knots. And they have some fears too, like, Sam Altman said in a recent interview that, you know, it’s okay to be scared of this. They too are a little bit scared of this.
And he said, you know, if, if I said I were not, you should either not trust me or be really unhappy that he’s got the job as CEO and I love that he said that. I love that because it was like, yeah, okay, great. You know, because there needs to be people who are leading the way with this, who are not just completely taking it for granted that all is sunshine and roses.
Kris: He didn’t brush off the question. He didn’t, he didn’t go like, she’ll be right mate.
Don: Yeah, it’ll be fine. Don’t worry about it.
Kris: Yeah, that’s right. Yeah. And, um, he even admitted there could be a set of bad, really bad outcomes. Yeah, and one of the things he mentioned was one of the big worries is that it could be used for large scale disinformation.
Yeah. Yeah. So, and when we’re in weird political times and all, you know, we know how dangerous that can be. And he’s also worried that the systems will be getting better at writing code and they could be used for cyber attacks and things like that. So, and, you know, We don’t know the evil minds who are like playing with this right now and, and thinking, Ooh, this is what I could do with this.
And because it’s such advanced, powerful technology. It’s impossible to predict.
Don: It is so impossible. Like we are sitting here getting excited about getting, you know, the AI to create, it’s a fluffy letter K in in fuchsia pink, and there are other people out there going, oh wow. I could actually do a cyber attack with this thing.
So there are so many people that make the world go around. And we don’t know who is going to get their hands on this technology and what they’re going to do with it. And because it is such incredibly advanced technology there’s a really, really good reason to be at least a little bit fearful.
Kris: Yeah. Yeah. So the things that we are going to talk about to be watchful of aren’t that sinister. No. Truth be told, we, we are keeping it to the level of graphic design business.
Don: Keeping it industry related. Yeah. Okay. Let’s, let’s leave the cyber attacks to the people who can actually manage that. Yes.
Kris: That’s not part of my skillset.
Don: No. Out of our wheelhouse. Let’s pull it back into design business management.
Kris: So, first one we wanted to talk about was the quality of the work and whether it is actually appropriate for the solution that you’re creating. Because like, you can do anything with this. Like you could ask these AI generators, the visual ones to create a cat in a boat wearing a rainbow sweater, drinking a cup of tea. Fabulous.
Don: Beautiful. Similar to the special fluffy pink K. Why? Why would you do that?
Kris: Yes. But would the result, be any good? For starters. So we are just looking at a lot of the AI artwork that’s been, you know, promoted out in the world at the moment and shared out in the world. And it’s got a style, right? It’s very complex. It’s got questionable aesthetics. Doesn’t it?
Don: It really has way too many elements, way too many layers. It’s got it all going on. It’s almost like the concepts are infinite.So let’s just pop them all into one thing. Yes, let’s just merge all those concepts together into one.
Kris: And it actually reminds me of when Photoshop first got layers and. All the designers went. Wow. And they’re layering everything up and adjusting the opacity and this element over this element. It was like just so crazy with the layered effects. And it was a real style for a bit. And was it any good? Not really, no. It really wasn’t. And did the aesthetic style last? No, no,
Don: No, it didn’t. Yeah, absolutely. And I think, it really does come down to the quote by, Clay Hebert. I really, I really just love that so much. I’m going to reference it again.
If you are average in our industry as graphic designers, as brand developers, as beautiful brand designers, if you are average, you are literally going to be toast because you’re not going to recognize when things are God awful in AI. You are just going to go, oh wow. Look at that. Look at all the things.
But if you are great at your job, you’re going to be even better. So, thanks, Clay. Love your quote. I’ll probably be sprouting that for a long time.
Kris: Yeah, yeah even when it comes to writing, it produces really complex stuff as well. So the visual seems complex. Like the writing’s complex, it’s repetitive, and it’s like, oh, you’ve just, you already said that. Now you’re just saying it in a different way and now I’m getting confused. It’s really robotic.
Don: Yeah. It really does lack that personality and, and brand voice, which, which we talked about in the last episode. If you do not know your brand voice, if you are not clear on your messaging, your business’ messaging, it’s time to step up into that.
And you really need to be clear on that before you enter this arena. Otherwise, AI can imprint a completely different brand voice over the top of you. So, so just be mindful of that. When, when it comes to writing.
Kris: Brand voice is incredible. It blows my mind, it reminds me of music sometimes because, you know, there’s just infinite possibilities with music and it’s like mathematical and we teach a process for developing a brand voice in the academy and it just is so amazing. There are no two brand voices the same, and we are teaching this process and you would think that they would be more similar, but all these beautiful designers, they’re brand voices are all different.
Don: Yeah, it’s amazing. Love it. We love seeing that. Yeah, absolutely.
Kris: So, if you are not clear on it, it’s going to spit out stuff that is not you.
Don: Exactly.
Kris: Not aligned to you, not in integrity with you. And you can tweak it and you can ask it to refine it to your brand voice, but you have to know what it is first.
Don: Yeah, absolutely. You need to know what you want it to do. You need to know where it’s going wrong. And if you can’t recognize where it’s wrong or where it lacks integrity or lacks depth or lacks emotion, then you’re not going to be able to input the right instruction to get the right answer. So remember, AI is only as good as the person inputting the instructions.
So, we can’t distill it down to anything more simple than that. Input equals output. And if you are a really good designer, if you are discerning, if you are elevated, if you have elevated systems in place, then the instructions you are likely to give AI are going to be elevated. And so therefore the output will be elevated.
But if you are just coming along going, design me a logo, my brand name is, is this, well then good luck. Goodness knows what AI’s going create for you. It’s going to be very subjective and we are always sprouting, be objective in the design process.
Kris: There’s this video from Wired where 20 people and their different jobs and it was, make McDonald’s logo vegan.
Don: Yes, that’s right.
Kris: It was hilarious. It was so bad.
Don: Yeah. We will link that again to this episode as well. But go back and listen to our rose colored version how AI is going to be wonderful in the first episode.
Kris: Yeah. So it’ll get it wrong. Like, it, it doesn’t always nail it, even when you ask it to refine it, you know, you can ask it, for example, make it more conversational, condense it, um, make it more loving.
We tried that, make it more loving because we just want to say, make it talk like Kris and Don. Why aren’t talking like Kris and Don? Um, and then when, when we said make it more loving, we had to say make it less cheesy.
Don: We still had to write it ourselves. So AI is not going to get our job anytime too soon either. But, you know, it’s, it’s provides a really good starting point and although it leaves us wondering if it takes just as long to fix the AI result as it would to create the thing from scratch.
Kris: I found that, yeah. Honestly, like it’s been exciting playing with it, but then I think that I’m molding it into shape and it’s taking me just as long,
Don: Way too long. Yeah, yeah. Way too long.
Kris: Because I don’t want it to not sound like us. No.
Don: It needs to be, that’s our brand voice and that’s that real clear understanding that we have about our messaging that we are not going to lose. Yeah. Because we, you know, we’ve worked really hard to, to get our audience to know who we are, and that is genuinely who we are, so we can’t lose that.
Kris: So if you put rubbish into AI, you’re going to get rubbish back. So it’s really only as good as what you input into it and it takes a lot of thought and strategy and intelligence and discernment to do that.
Don: It really does. Yeah. It really does. And it’s fun though. It’s fun. But what we recommend as coaches who teach designers how to systemize their business, we also recommend that you start systemizing your input into AI as well. So get yourself a set of questions that you ask on repeat and have them on a, we use Trello in our business. We have beautiful Trello lists and Trello cards.
So we recommend getting a bunch of questions that you would ask through the briefing process. Get a bunch of questions that you would ask, you know, when you are putting together a, a plan, your social media so that you’ve got them ready to go, because you don’t want to be trying to remember what you say all the time when you’re inputting that into AI.
So start to systemize this. Make sure you get on top of that. So a little pro tip there.
Kris: There’s so many different ways people are asking AI, and if you watch a few videos and you see people convert text and, and get a better result, just take note of the commands that they’re giving. Yeah. Because you’ll go, oh, that’s an interesting way to phrase it and that’s getting a good result. Like, we had fun last night. you gave me the example where you said make it younger. Yeah. Make it younger. And so we did one with which was make it sound like a 13 year old girl. And, and it’s really funny. There’s examples out there, like make it sound like Snoop Dogg, make it sound like Shakespeare, you can do so much.
Don: And then you can actually have those as a voice, so you can set them on the video that we’ll also link to today’s episode. They had, Oprah. And when you are asking for advice, put it in the way that Oprah would say this. I mean, that was just gold. You can have so much fun. You can, I really do believe you can have lots of fun with this, but isn’t this meant to be our dark side? So we’re going back into the joy of it, Kris, because we’re having so much fun with it.
So anyway, but going to pop our serious hats back on and let’s have a look at the next thing that we want you to question and to think about. So the next thing is ethics. So this technology is taking references from many different sources to create something brand new, but where is it sourcing it from?
It feels a little bit like the wild, wild west at the moment when it comes to appropriating other people’s work and the ethical dilemma that comes with that.
That is a really big thing for Kris and I. Who was the original generator of that thing? And why do I get to have that now? How does that work? So this seems especially problematic for AI visual generators, so it’s really ethically ambiguous.
Kris: Yeah. Where are these apps scraping their source material from? Where are they getting it from?
Don: Yep.
Kris: And there’s actually some lawsuits going on already.
Don: Already! And it’s been months, like it’s only been months that this has been going on. So already there’s some lawsuits.
Kris: Yeah. Like Getty Images. Getty Images is, if you don’t know Getty, they’re like the leaders in stock photography, they own iStock photo. They are suing the creators of an AI. Art software, called Stability AI, and apparently they have been scraping some of Getty’s content. Allegedly.
Don: It’s all alleged. They’re in the throes of that at the moment. So we, we can’t really speak to that. We don’t really know what’s going on, but we have heard on the grapevine, like all other businesses that that’s what’s happening and so we are really shocked to find that that was happening already.
Kris: Yeah. And um, Adobe is addressing it. If you look at their website and have a look at, um, what they’re saying about ethics and copyright, they have, they’re trying to address it, but it’s a little bit gray and you tell that they’re sort of throwing it together at the moment. You just get this impression that nobody knows what they’re doing.
Don: Nobody knows, really. Flying by the seat of their pants. Literally. That’s why it’s called Firefly.
Kris: Yeah, Adobe’s offering. It’s called Firefly.
Don: Yeah.
Kris: So apparently that one is trained on Adobe stock images. So they would have licensing rights to that, yep.
Don: Which is good. Tick.
Kris: Then they say openly licensed content. Tick. Okay. And public domain content where copyright has expired. But it concerns me a little bit because openly licensed content, what if somebody decides, I don’t want it to be publicly licensed anymore.
I retract take that back and you’ve created something in, I don’t know, it seems a bit, it seems fishy to me and maybe Adobe’s going to be quite a high standard of ethics. Yeah. But some of these other ones, I’m not so sure we, we’ll see what happens.
Don: True, it’s just so gray, but also from Adobe that they have said that they’re developing a compensation model for Adobe stock contributors, which is really nice.
And they’ve said, we’ll share the details of this model when Firefly exits beta. So it’s in beta at the moment. And as soon as that happens, Adobe have promised a compensation model, which I think is really great.
Kris: Yeah, that’s something, yeah, isn’t it? Yeah. So they’re really, they are thinking about it, but these waters are very muddied at the moment. So we strongly recommend, and we’re not legal professionals, but this is our gut feeling here. Don’t use AI visuals, AI generative visuals for any published work for clients or for yourself until this is sorted out.
Don: And clear. We want real clarity around this because we want to have complete transparency when it comes to AI and how we use AI in our design businesses.
So until then, brainstorming is perfectly fine. Conceptual purposes is perfectly fine. You know, creating a catalyst for your own thought is perfectly fine. So just think about it when you are about to use it. Am I in integrity? Is there any gray area here? And make sure that you feel really, really good about what you are producing.
So we really do recommend steer clear just for the minute for putting it out there as published works for your clients or yourself because it will get clearer. It’s, it’s just in the, you know, the design squiggle. It’s in the design squiggle at the moment. And it will start to refine and it will get clearer down the track.
Kris: Yeah. It’s a big muddle. So use it for brainstorming. Use it for conceptual development, but don’t share it publicly. Don’t publish it.
Don: And we believe it won’t be long, it’s happening so fast that you won’t have to wait long before you’ll know the parameters and how you can legally use it.
Kris: So another concern that we have is that there seems to be a communication bias with the written forms especially. Yeah. So, racial bias, a gender bias, and it’s quite rigid, as we mentioned before, bit robotic and dare we say it, it’s a bit mansplainy.
Don: Yes.
Kris: So it’s currently based on what the masses, so the masses in inverted commas, the masses have created pre 2021. And this was interesting because my daughter asked about something that’s really only fairly current, like it’s a 2022 thing, or she wanted, um, to expand upon something to see how it worked.
And I said to her, I don’t know if it’ll pick up on that because it’s based on 2021 and prior data, and so this is, that’s too new. But who says the masses are right in their thinking? Yeah. So minority groups and people who are part of marginalized groups in society. Like, is their voice going to be heard?
Is it’s supposed to be original generative content, right? But is it really producing relevant and exciting new thoughts that are actually groundbreaking?
Don: Yeah, absolutely and again, I think we come back to what is being input reflects what’s being output. So if we are finding that there’s a racial bias or a gender bias, or it’s rigid or whatever, ask the question differently, frame it differently. Until you start to educate because remember we can actually tell ChatGPT when it’s wrong. We can actually say what we’d like to see. So the more people in the minorities that actually speak to it will actually help to educate and inform it Um, because at the moment we’re not seeing that enough. We’re not seeing that.
Kris: Like we could say no, I don’t like that make it non-binary or no, I don’t like that make it more inclusive. So it’s, it’s something that we have to train it to do and to, um, get smarter and smarter at that as well.
Don: Yeah, absolutely the next thing that we want to chat with you about is copyright and ownership. So the legal rights, you know, who owns it, who owns the copyright. So if you have created a logo, using AI technology for a client. This is likely to be owned by the platform you created it on, not you, not the client who’s buying it from you, but the platform that it was created on.
So again, this really seems like a very gray area, and it could be a copyright shit show.
Kris: Yes.
Don:cOr is copyright and intellectual property rights going to be a thing of the past?
Kris: I feel like, I don’t know, it’s sort of unraveling somehow. Like I think about the new generations coming through and how much they repurpose other content. Like my daughter who’s 14, she loves editing.
So she will do all these edits of her favorite show or her favorite movie. And it’s all this materials that you find, right? Oh, on online that probably somebody else got illegally. And like, I just feel like, remember when music first started being able to be downloadable? Yeah. And it was, it was so big.
And all the record companies fought against it and it was like, no, this can’t happen. But then technology adjusted to that. And I think when there is such an open platform, how is it going to be regulated? How is it going to be protected? So I’m not sure if like, who knows how it’s going to go, but it feels like it’s being, you know, that we mentioned this on the last podcast, but disruption tends to happen when there’s a new technology like this.
Don: Yeah. It’s a massive disruption.
Kris: Who knows?
Don: Who knows? So this is what Adobe is saying about content creating, using AI and no doubt other platforms will do something really similar, but what they are saying is Firefly will automatically attach a tag in the embedded content credentials to make AI generated art and content easily distinguishable from works created without generative AI.
Okay, there you go. They’re doing it already. They’ve already come up with a bit of a solution. Yeah. So thank you Adobe for that.
Kris: Yeah. And Google has a bit to say about it as well, and because people have been concerned if I use AI generated content for a blog, for example, will it still be picked up by Google? Will Google prioritize it with SEO and um, with search results and all that sort of thing? And Google has essentially said it’s fine to use AI, but it still has to be relevant. There’s all these criteria that Google decides that makes something relevant or not, and makes it worth promoting or not essentially.
So one of the major ones was being relevant for humans and so if you have this robotic text, it’s not really going to come across as relevant for humans.
Don: Yeah. So Google will ignore it. Pretty much. So be very, very careful. Be mindful again, know your voice, know your brand voice design business owners out there.
All right. Bringing us to our final point, accuracy. So this is one thing that really scares us a lot and Kris and I have had a play this past week and we have noticed there’s a lot of inaccuracy. We might go into an AI platform and we will get information back whether AI model that we’ve used, we’ll confidently state things as facts. They’ll, they’ll be like, these are the facts and they’re completely wrong.
It’s like, oh my goodness. Here’s what ChatGP T says when you land on their main page. This is great. May occasionally generate incorrect information. Okay. Just beware. May occasionally produce harmful instructions or biased content. Oh, limited knowledge of world and events after 2021.
Kris: Alrighty. Yeah. Okay. So, so limitations, right? And it is interesting because you know, you could start using it like people have been calling it Dr G P T and you know, you could use it for so many different things. you could ask it to diagnose some health problems or something. And so that could be really factually incorrect.
Don: Yeah, yeah.
Kris: You know, diagnose this rash, it looks like this, you’re describing it.
Don: And again, output is only as good as the input. So what if your ability to describe things is not so good? You know, like, for example, I’m going to throw my beautiful husband under the bus. I wear a lot of black, I love black designer black. I know, I’m sorry, cliche. However, I have quite a few different black tops, and I’ll get a new black top and he’ll go, oh, look, what was wrong with the other one? What? What’s wrong with the, the last one that you got? Oh, it’s so different. I’ll say. It’s completely different. It’s a different sleeve. It’s a different cut, it’s a different fabric.
He doesn’t see the subtle nuances in the design of the top and so therefore, I would be very scared if Dean was to describe a rash.
Kris: Yes.
Don: If he can’t describe a different sleeve, how is he going to describe a rash?
Kris: Do you know when you first started talking about this story and you said, I’m going to throw my husband under the bus here, I thought you were going to describe a medical problem.
Don: He’s totally fine, but he cannot describe things very well. So he’s not a visual descriptive person. So if he was to use ChatGPT, we’d be in all sorts of trouble. We’ll have all sorts of ailments that we don’t have, so be very mindful. That scares us a little bit.
Kris: Well, yeah. And you know, you know what they say about don’t, Dr. Google will probably don’t Dr. GPT as well.
Don: Yeah, absolutely. Absolutely. So in conclusion, beautiful people. So add AI to you tool kit.
We absolutely believe that. It’s a big tick from us. We are going to be playing with this for now and forever, I’m sure. Have a play, and learn how to use the tool to the best of its abilities and make sure that you are pushing the boundaries, that you are in control, that you are the leader and the guide. So, yeah.
Kris: And just know that no one is an expert at this yet.
Don: No, we’re all in this together. We’re all in the deep
Kris: together. You’re not lagging behind. So if you haven’t even opened up one of the AI platforms yet, or you haven’t logged into ChatGPT you’re not lagging behind, just, you know, have a play.
Don: Yeah.
Kris: You know, as we mentioned in the beginning of this episode and in, and in the beginning of the last episode as well, this is such fast moving technology that no doubt this won’t be the last time we’re talking about this.
Don: No, there’ll be so many nuances that we’ll want to unpack with our audience.
we’re sure to have lots more conversation.
Kris: And if you have any questions, we will love to hear them because as we talk about this more, we might look back on this episode and have a little chuckle about how naive we were the beginning.
Don: Yeah. Literally.cWe’ll have a laugh about this definitely in the future because, you know, it is so, so brand new and it’s okay. It’s okay to be in this together and that we’re all in the deep end together.
Kris: SocWe would be fascinated to hear your thoughts around this. Do you think it’s a good idea?
Don: Yeah. Are you worried about it?
Kris: Yeah. Yeah.
Don: Let us know and send us a DM over on Instagram and let’s chat.cSo at Design and Prosper on Instagram. All right, beautiful ones. Don’t bury your head in the sand.
Kris: Definitely don’t do that.
Don: Stay informed and don’t be afraid of engaging in the conversation. Knowledge is power. Really is. Yeah. Yeah.
Kris: Yeah. Okay, we’ll leave it there and we’ll be back very soon.
Don: Yeah.
Kris: Have a beautiful day everybody. Bye. Bye.
Don: Bye.
© Copyright 2024 Design & Prosper Pty Ltd