How Does AI Change the Landscape for Brands?


Graphic of a robotic hand reaching out of the screen of a laptop to touch its finger to a photo of a human hand , with red concentric circles at the point where they touch.
Solving for B°
How Does AI Change the Landscape for Brands?

Artificial intelligence is the latest hot topic in nearly every industry, which means there are all sorts of exciting opportunities to innovate in branding and marketing. In this episode of the podcast, Chris is joined by BrandExtract VP of Technology Donovan Buck, Chairman Jonathan Fisher and CEO Bo Bothe to discuss some of the ways AI may shake up the market for brands and what their top priorities should be.

Table of Contents

This transcript has been edited for readability.

What is Artificial Intelligence?

Chris Wilks: AI is the hottest topic out there right now and we thought it might be helpful to talk about its impact on the branding and marketing world. So Donovan, to start us off, can you tell us a little bit about what AI is and what are we referring to when we talk about artificial intelligence?

Donovan Buck: Sure. Well, the term itself has been around a really long time, since I was a kid, and it's meaning has evolved. Back then, it was having clever sub-routines that maybe processed the data in new and clever ways.

But what it's come to mean lately is training systems on large volumes of data and then creating responses based on the data that's been consumed. That's been enabled by:

  • One, the large volume of data that's available digitally now
  • and two, the ability to process more and process more in a timely manner has made it possible to do these things where we're just consuming huge amounts of information and then using that to inform the most likely responses.

Chris: Is there a reason why AI has kind of taken off like it has in recent times? Is there a big jump in computing power? Is there a catalyst for this or is it just luck?

Donovan: I think some of the new models, particularly those that present themselves as generative tools (the things that are creating new things from seemingly nothing), their capabilities have gotten so good recently that it's captured a lot of people's imagination about what's possible and what's coming, and for good reason.

The things that large language learning models, things like ChatGPT, can do are kind of mind-blowing. Even though they're not perfect, they're really good and I think that's just generated a lot of buzz around it and a lot of momentum where a lot of organizations are offering tools that use these things to enhance their own product offerings.

Bo Bothe: I think effectively, Chris, it's become more commercial... more commercially viable, where someone like Donovan could have used tools like this before to bend code and bend things, computer processing power to their will. Or the masses have access to it and then they start to have ideas and then those ideas turn into other ideas and people can build on it.

And Donovan's talked about this in the past where these platforms are... Some of them are buildable on or usable in different ways through what you've talked about APIs and other things, that it just makes it way more accessible to a larger number of people, and you're asking, are we at an inflection point?

We're probably at the "oh, cool" inflection point, and then at some point we're going to get to the completely commercializable inflection point, but time will tell and that's happening faster and faster in our lifetimes as things make us say, "Oh, that's cool and new. Wait, here's the next new thing," and that's happening faster and faster now.

How Can Brands Use AI?

Chris: So what sort of movement are we seeing right now? Donovan, You talked about generative AI, which is seemingly pulling things out of midair. I've worked with ChatGPT, DALL-E, Anyword, Bard, Midjourney. There's all these different tools. So what are some of the ways that it's really manifesting itself that's relevant to brands and marketing teams and brand firms?

Donovan: So it's interesting because I think each discipline, in our business for example, has found different ways to leverage these tools, whether it be something that's specifically optimized for doing a single task or having something broader like ChatGPT that we can leverage and do other things with.

So you almost have to say, "In this discipline, what are we doing?" For writers, what's the added value? For developers, what's the added value? Because it's a little different for everybody and the capabilities are different for everybody.

Chris: So let me ask you, D, as a developer, what are some of the things that developers can use this for? Even if you're new to AI and you're like, "Hey, look, this can really help you do X or this can really help you do Y. This is where you might want to start." Do you have any ideas on that?

Well, the obvious one is GitHub Copilot, which is... GitHub's a Microsoft company. And it's been around a little over a year and we were fortunate we got into some of the betas and early releases to start testing it out for our development processes. And what it does is, it's like an autocomplete for code.

As you're typing your code, maybe you give it a prompt, maybe you give it a method signature, you say what the method's going to look like, and it will start to write the next line for you, and if you like what it wrote, you can just tab through it, just like if you're using Grammarly or something like that where it can offer suggestions on what you might want to say next. 

I think GitHub announced a few months ago that they were up to a 37% acceptance rate on the code that they generate for developers, and that's pretty impressive. Everyone on our team, even when we have skeptics and people who are reluctant to take on new things, has said it has helped them and helped them be more productive.

Things like code are very formulaic. Well, it either works or it doesn't. There's different ways to accomplish a task, but for the most part, it either works or it doesn't. It is very structured what you have to create, so it's a good fit for what AI can do. Sometimes you will get nonsense, and that's where your training kicks in and you know how to kick that back out and say, "This is not good code."

Chris: I'm curious, Jonathan, from your perspective, we talked about the uses of AI and it doesn't necessarily have to be a tool, but I'd be curious to know from your perspective, what are some ways that brands or marketing departments can use AI today?

Jonathan Fisher: Well, they're using it, for example, with programmatic advertising and media buy. AIs do not need to develop personas to the level that a traditional marketer does because they can go out and identify behavior and then find like behaviors from other potential prospects in the marketplace. So personalization is another way that the marketers are using it today.

Instead of having a writer write one thing and it's kind of generic, the AIs can write specific to those behaviors and those buyer patterns that it's seeing in the process so they can produce more content and distribute it in broader, more unique ways in the marketplace.

We're using it in research analysis, whether it's through voice of the customer interviews to help analyze the data from the customer interview itself to identifying trends within the data itself or the insights, positive or negative, verbatims and comments, sentiment monitoring and management. It's being used for that.

Those are a few of the, I think, more adopted already methodologies that are out there. Bo's got a few I know that he's seen as well. 

Bo: Thanks, Jonathan. I think the big thing is, it's really a time saver at this point. There is a creativity saver in that, based on what I prompt and based on what it can see, or what these different tools can see, it can feed me options faster and quicker. And I think much like the code example, much like Jonathan's talking about on looking at research and get to the next conclusion, get to the next conclusion.

So I think it's a tool that's going to allow us to be... all the companies and even our industry... is going to allow us to maybe speed up some things to get to a certain level of basic or push beyond and give us some ideas we wouldn't have thought about as quickly and then we can build on it. I think, Donovan, we've all talked about this, how quickly these different tools start to learn from each other and learn from what we give it for it to be able to make more kinds of things.

In Midjourney, I tried to create a picture of my wife as an elf in a library because she's a librarian and she geeked out to the whole Lord of the Rings thing and I just couldn't get it there, and I think it was really more of a problem with me and my prompts and my ability to use the tool than it was the tool itself, but it still needed information, it still needed guidance, it still needed more and no matter how many times I hit refresh to give me a new option, it made her look older or younger or whatever.

Those kinds of things, playing with those tools, it's interesting to see where that is. I will say there are going to be industries and there are going to be professions that are almost all significantly disrupted or going to go away.

That doesn't mean that everybody should run out and start cutting things because the robots are going to take it over, but at the end of the day, we should be thinking we've gone through this with the computer, we've gone through this with the handheld devices, we've gone through this already, and industries have changed and I think we're at that point.

It's just that the challenge for us now is that it's happening so fast and like Donovan talked about, the computing power is getting better and better and better and people are getting more and more used to using these tools and that's going to affect the product you sell, that's going to affect the quality of your product, that's going to affect your people's psyche when you start talking about branding and believing in a brand. That's going to make people question, did a robot make this or did a person make this?

What's the value of it if it was made super fast? The mentality of, "I value the pen to paper," or, "I value the sewing and the stitching," it still permeates business and in some cases, the robot should be stitching. The robot should be drawing. How do you navigate that or how do you upskill your people and your team to be able to meet that challenge or meet the tool and use the tool the way it can be?

I think those are the things that business owners and business leaders should be thinking about and that's going to affect trust in brands, that's going to affect hiring, that's going to affect people who are going to be afraid for their jobs. How do you communicate your strategy around these kinds of things? All of that stuff are things that at least our clients and we are thinking about when we talk about these things.

Will AI Disrupt the Workforce?

Chris: So it's interesting when we talk about potentially upskilling and how this will disrupt different industries because I've had conversations with folks on our team who might be writers or might be designers and there's maybe a little bit of panic and a little bit of curiosity of like, "Hey, look, what's this going to do to me? What's my career going to look like in the next 5, 10, 15, 20 years?"

And I think you're right, but I think it's absolutely going to disrupt and it's going to change things, but what I believe is that the folks that will be most impacted by this are the folks that put their heads in the sand and say, "Well, this is a fad," or, "I don't need to change the way I do things."

The people who will be successful in this new era of marketing and branding, if you will, are the people who embrace AI and understand how to use it. Understand its shortcomings, understanding its pitfalls, understanding that it is rarely, at least not at this point, going to spit out a final product. It'll help you with writer's block. It'll help you generate some ideas.

Maybe if you chat with it, if you use the ChatGPT for example, it can probably help you come up with some good suggestions and ideas, but I think the value still remains in the human where it's on us to, pardon the pun, extract that value and extract the really good idea and take it to that next level. And I'm curious if you guys see it the same way or if you have a different perspective on that. I'd almost want to direct that comment or that question or comment at Donovan because I think you among the four of us probably understand the power of AI to a better degree.

Yeah, I think you're spot on. Say you do want to use ChatGPT for brainstorming and you say, "Give me 20 ideas for a name for a business that does X and Y," and it spits out 19 horrible ideas and one that's pretty good. You've got to be able to discern it and you've got to be able to take that idea, which is just a seed, and grow it into something that has value.

I see people, sometimes they'll start with ChatGPT and they'll give it a prompt that's relatively complex and it will return something that's impressive but fundamentally flawed in some way. It doesn't capture the voice right, it doesn't make sense at parts, but when you learn how to work with the tools and really massage the information that you get back and how to create a really good prompt to help inform that tool how best to do its job, then you can start really realizing some great benefits.

If you're not doing that, then it's just a toy and you're not even getting back Fiverr quality, if you were to hire a writer on Fiverr for example. But there's some really great resources that help you understand how to engineer the prompts and use the tools. When I talk about prompts and stuff, obviously I'm thinking of the ChatGPTs of the world, not the off-the-shelf tools that are meant to help you do your job, but that's a way that you can create your own tools from those basic products that are out there and build upon things that we couldn't necessarily build in-house.

Balancing AI With Original Ideas

Chris: I remember one of you guys mentioning that all of the fodder, or the learning, comes from a human. Something to the effect of: AI can't quite imagine something that wasn't a seedling of an idea before or something to that effect. Can you guys talk about the need for the seedling idea? 

Donovan: That was Bo. He said quite brilliantly, I thought, that AIs are not good at imagining things that haven't been imagined before. They're consuming these corpses of text and if something hasn't been written before, it's not going to stumble across and write that itself. So yeah, I thought that was a really brilliant point.

Bo: On the other side of that, Jonathan and I debated it in that it might be based on a prompt imagining something that I couldn't have imagined. But to your point, D, it's based on my ideation, my ability to put the right words together with things that I think are out there, styles or whatever, or writing styles that can then mimic something the way I want it to mimic it.

And I think Jonathan brought this up and it was a really good debate or debate point yesterday... The naming experiments that he's running that I've run, there are 10 good ones out of the 30 it gives you if you give it the right information.

And that doesn't mean they're all viable, that they're all copyrightable, that they're all trademarkable, but you can start to iterate, iterate, iterate a lot faster, which I know Jonathan loves and I'll turn it over to him, but Jonathan loves that volume of ideas so that we can get to the right one. Again, the human being still needs to make the decision as to, is this unique, new and awesome and does it fit? But the robot can give you a volume.

Jonathan: Yeah, I don't think it's sentient just yet. AI is creating things that we may imagine but can't produce. It can run scenarios 10,000 times faster than we can, so the possibilities become more endless for it than they are for the human individual to work through this and the options that are out there.

So for researching, for collecting data and all that, spot on, but back to the earlier comments, what are the risks? If it's garbage in, it's garbage out. If it's only 30% right, it's only 30% right, so you still have that expert individual's analysis and insights on top of these tools, but these tools foster efficiencies and they facilitate actions that are maybe not even possible for some of us in our professions and our careers at this point.

And to Bo's point earlier: the timetable by which these things are adopted now is just incredible. It was 50 years roughly for the car to replace the horse and 20 years for the computer to replace the manual calculus, but then you start looking at the adoption of the phone, the smartphone's less than 10 years, and now they're talking probably five more years of mass consumption on these AI products in the marketplace before their true mainstream and norm. So that blur between bleeding edge and leading edge is becoming very, very thin.

Bo: Yeah, but I think that's... When you start to think about the impact on brands, it really comes down to if these tools allow us, people like us, to create things that haven't been imagined or been produced before, then those things will be consumed by AI and it'll be this constant cycle of one-ups-manship that you have to stay away from.

We've always tried to tell our clients that, like, "Hammer away." Especially in the B2B world, not enough people are going to see it right away. You might get bored with it and AI might be able to create something new tomorrow, but you still got to hammer away your message and your story, but the fact that... I think, Chris, what Jonathan and D have been talking about is that while the tool is here right now, it's going to be here so much faster than previous technologies.

We used typewriters before we really started to adopt computing power to help us do more than just type setting, and some of us bent that tool to our will a little faster. It's the same thing with the mobile phone and the handheld. We used it as a phone for a long time, but then it basically became our lives where we consumed information and we were able to share information.

That happens so much faster than the computer. You have to look at this the same way to Jonathan's point. 5 years, if people can explain their idea, something can create it faster than pen to paper used to be able to. Donovan worked in pre-press and worked in a film house that output film to go plate things. That whole industry's gone now and it's included in a tiny little box in a machine that spits the film out. That's mind-blowing to someone like us that started where we were, hand cutting Ruby to take a picture of to burn it on film.

Now we don't even have film. It's just direct to print. Whole industries were gone, but we still were able to create new ones and new ways to do things. Donovan got into code and development and created a new way to share information rather than the old way of film and print.

And so those are the things that I think business leaders need to really be thinking about. Those are things we need to be thinking about and young people going out into the world, how am I going to use these tools to get my ideas out of my head to create something new? It's just going to be faster and quicker if it's well thought-out.

Donovan: I was just going to add on that yes, this wave is crashing upon our business and everyone's business for that matter. There is going to be some pushback. There's going to be a point where people start to realize the limitations and the things that it does not do well, we talked about earlier.

There's going to be more news stories about lawyers who submitted briefs that ChatGPT wrote for them and they were just completely wrong. There's going to be doctors misdiagnosing people. Bad things are going to happen because people are trying to use the tool in ways that it's not capable of yet, and there's going to be this pushback against AI and it's coming.

But businesses that continue to work with it and figure out how it can improve their operations, improve their products, improve the way that they deal with their customers are going to continue to reap the benefits and it's going to become part of our lives just like the computer and the smartphone have.

What are the Limitations of AI?

Chris: Yeah, and you touched on something that I was getting to, which was some of the limitations that you're aware of with AI right now. For me, for example, you guys touched on this already, but I'm noticing in copy, when it first came out, it was so novel and you were all excited about it. "Hey, it spit out this copy. This is great." And then you look into it and then you realize what it's missing.

You realize that, "Hey, it doesn't quite have this right or it doesn't quite understand this." So I think there's some limitations. Are there any other specific limitations that you guys watch out for when you're using AI? Just to be careful because you don't want those lawsuits or those bad medical outcomes and that sort of thing that could be a real possibility.

Well, I was going to say with programmatic automation and marketing buys, it's going to go rogue on you and you're going to have some waste in the buy before you can catch it and their trends are going to put up, but it's also to say, do humans make perfect buys too? No, they don't. So there's a little bit more of an agile mindset that I think you have to take with some of this process in there.

So intellectual property, is it robbing copy from somebody else or stealing visuals? There's a whole lot of IP rights right now around what DALL-E can do and creating visuals and stuff out there that is... The courts haven't even thought to consider or figure out, but as a brand and as a marketer, I think you're going to have this challenge of how do you defend and protect your brand and/or how do you keep your people from... deviating so far from the brand that you can't protect it that way either.

So there's the brand policing process and there's the monitoring process of it, and then there's just the fake communications and deep fakes and news that's out there. There's another form of protection that brands, if you talk about risks, that you're going to have to be monitoring and looking for. So it brings up a lot more complexity.

Our industry was already getting more complex every day. You have to go from the internet to then social media and pre-that, desktop posting and every year, there seems to be some new thing as a marketer that we have to contend with; a new app, a new platform that comes out. 

Chris: Yeah. It occurs to me that with this new dawning of generative AI, I think the people who get a handle on AI first and embrace it will probably be in the best position to succeed, but really being nimble and agile and being adaptable to all these new technologies that are going to come out.

Because Jonathan, as you were talking about that, you think about bad actors and ways that they might try to sabotage their competition and put out a deep fake video about them and their stock price tanks.

And if you don't become comfortable with it, you won't know how to even combat or be prepared to have your brand story buttoned up and have built up some reputation for, "Hey, look, we can prove that this isn't our CEO doing X, Y, Z, but right now, our brand stands for itself. While we're sorting through the technical details of that, let's stop the stock price from plummeting."

There's just so much that even in this conversation, I'm like, yep, didn't consider that. It is a new wild, wild west, it seems like.

Jonathan: I think anybody that deals in crisis communications is going to be very, very busy.

Donovan: Well, there's multiple facets to this problem. As individuals, as consumers of information, the necessity for critical thinking is going to go up as the quality of the fakes go up. It's like, "That seems really unlikely." People sometimes seem a little too quick to believe something when they see it without questioning it.

We're going to have to all get better at questioning the things that we see. As producers, that's a really good question. You think like a watermark, how do you protect that image? How do you protect that copy? How do you protect your brand image? Just your own face? Do you watermark your face? How's that work?

And then there's the regulatory facet. Some governments will. I don't know which ones, but some governments will come out and give guidelines about appropriate use of AI and AI-generated content. Does it have to have a little legal snippet at the bottom and that's going to happen in some places? Whether it happens here, I don't know, but it's something that we're going to have to consider in the work that we do.

Bo: This is going to be lost on younger people that listen to this: we used to have to imagine a story about Egypt, and we would go to the library and we would open up books and look at articles and we would do the imagining in the front and the research in the front and then off of literally two chapters and five paragraphs out of the World book, we'd write a three-page paper on Egypt. And I had no business writing that, but I had to use my imagination.

I think people are moving so fast. Information is so readily available that to Donovan's point about this double checking and critical thinking, we're going to have to do the research on the backside of these now, and that the burden has shifted from where I was the content creator in that scenario in high school writing a paper on the pyramids in Egypt, the burden's now shifting to the reader and the consumer of the information to check and see if it's accurate.

It's almost like we're having to think backwards from the "imagine, research, write" process. Now it's going to be a "write, reimagine, research" kind of thing. It's a real change in the way that people do work. And again, brands are going to have to think about that. They're going to have to think about what training do people need. How do we get that and flip it back to well-researched information provided?

I think we're going to have to be a lot more curious and that curiosity going into using the tools the right way will separate the wheat from the chaff. The really good work that comes in part by using AI will be about really good people, really smart, intuitive people coming up with solutions and it's still a people driven process that is, to Jonathan's point, spitting out 10 of 10 really good ideas really fast to get to the next 10 because the URL's already taken on a naming project or that logo's already been done.

All business leaders are going to have to consider separate disclaimers that a government makes you do or a legal issue that comes from copyright. We are the creators. We are the imaginers. We're just using that new pair of hands to create something a little faster and a little bit quicker than we wanted to and to imagine some things for us that we don't have the time to imagine on our own.

Jonathan: One of the things that Bo talks about here in his explanation there is this notion of changing behavior and shifting behavior and not only to who but to when. And so if you take something as basic as Google search or Bing for SEO and organic searches, how the searches are conducted and how AI will conduct the searches is different than today's approach for the most part.

And so it's more who, what, how, where, when, why line of questioning from a content creation standpoint than it was just the old data and the facts and the stats and then the product and service explanations. And so something that's already on top of us because of the adoption curve that's coming for things like search and website and social media, we as marketers have to think about how we're delivering the data differently so that it can be consumed under these new methodologies and time periods by which it is consumed under. That customer journey is changing a bit, to Bo's point.

Donovan: Well, that's another aspect we haven't really discussed so far, and that's the relationship between consumers and data aggregators and what data aggregators are going to do with these new capabilities.

I'll use Meta because they're the one everyone knows, but there's some that are collecting a lot more information that we've never heard of that aggregate data and share it with different organizations to help them better target you as a consumer. What's going to happen to our personal information now that we've unleashed AI in the world? It's scary to be honest.

Top Tips for Brands Using AI

Chris: So let's bring it all the way back now: if a brand is considering using AI or is starting to use AI, what sort of advice would you guys say to brands out there who are trying to make sense of this and want to use it for their brand?

Jonathan: Well, I think that whatever tools you're using and picking, you want to make sure that they align with your values and your beliefs and your opinions because the AI isn't going to know that out of the gate. And we're not quite there yet in terms of it being able to pick up on it or even train it. You can teach it voice, but you can't teach it values necessarily. It's still a human characteristic.

So I would really caution that you're not letting it run on its own and/or trusting it too far if you're running some multivariate, A/B, agile type testing processes that are out there. I think additionally, brands are probably going to need to be a little concerned with trust and loyalty from customers and the potential backlash if these things go rogue or blow up on them. And so they need to consider a disclosure, how and when and what they're using it for and how honest they are behind it.

We've seen what happens when brands get caught lying and that is the cardinal sin of branding, is breaking your promise and that consistency is the golden rule, if you will, to all marketers. So really think twice about it.

And the third thing I'd say is really watch your IP issues. And I would be very careful both in protecting it and using it to generate content and/or generate visuals because I think there's going to be some pushback and there's going to be people testing the courts.

We saw it just in things a few years ago with sampling and with record in music that's gone on, but we've also seen it in just art, in the art world itself, the images of Michael Jackson and Prince being used and not being modified enough and things like that. So those are two things that I would be mindful.

Donovan: I've got two things I want to touch on. The first is as you're deciding what tools to use in your organization, think about your existing processes and how those tools can fit in and augment those processes.

Don't just buy things because they're impressive or you sign a license or start trying to use them without knowing how they're going to improve your work product, make you more efficient, or in some way help your organization as it exists now. And don't just make a list of tools and say, "I want to buy that, that, that and that," because you'll just waste your money.

And the second thing that I want to say from our own experiences in learning, because we're doing a lot of work with ChatGPT right now, is the... Don't underestimate the value of training the model you're working with.

For example, even with what's called few-shot learning, where you're just giving it a couple of examples and showing it what you want it to look like and then asking it to produce something for you. That dramatically increases the quality of what you receive.

For example, we asked ChatGPT to write a description of a webpage to use in a meta field, a meta description for SEO, and it did a nice job, but it was too long. It didn't use the voice that we wanted.

It missed some important points, but then we tried the same task saying, "Here's a few that we've written for this website, for these kinds of pages. Now you try it," and it returns something that was incredibly good. So know that you have to train the models. They don't come out of the box ideal for your situation. You got to tell them what to do.

Bo: Yeah. I think, building on that, are you generating ideas? This is not final art. You've got to put a little bit of eyeball on it and test it and look at it and we're talking content, but I would say the same thing about automating processes.

I would say the same thing about if we're automating media buys, it's going to get so far out in front of us if we're not watching it, that it may repeat the same mistake over and over faster than would've had if human beings had been just paying attention to it. Or human beings had just been slowly doing what they had done before. And so the mistakes can get bigger faster if you don't think about these tools as idea generation and not final.

It may give you something final right off the bat. It may give you something perfect, but you've got to be curious about that. And I think the last piece is, then, what kind of disruption does that cause in your organization, whether it's your own people trusting the business, whether it's the... "Is my job going to be here?"

When a business leader comes out and says, "Our brand believes in this, but we're going to have all the robots replace everybody," that's what goes on in their minds and how do you keep your culture intact? How do you manage this stuff the right way?

Jonathan: Treat AI like a new hire at your company. You're going to be teaching it a lot about your business and bringing it up to speed and review them just the way you would a new employee. That's how I think of AI personally right now.

Donovan: Yeah, it's that eager intern who has no idea what they're doing.

Chris: So this was great, guys. I really appreciate the time. It's really insightful. I learned a few things today, but yeah, we'll let you run and thanks a lot for joining us.