Episode 22

Will AI Kill Jobs or Create Better Ones? Business Owners Debate.

Key Takeaways

  • James Thornton says his company’s entire customer service and quality assurance group is now powered by AI solutions, which resolve customer concerns quickly, though AI does not get you 100% of the way there and the human touch still matters.
  • Cliff Nonnenmacher uses AI for decision support, content, and operational leverage. He says a 10% net margin business that cuts its price by 5% must double in size to break even, and that moving from a $5,000 a month content creator to a $20 a month AI subscription saves about $60,000 a year.
  • Eric Kanagy works in critical infrastructure, where data sits locked in arcane vendor interfaces. With an AI prompt, users can now generate dashboards themselves, and one person who spent a week a month on water usage reports can produce them in seconds.
  • Richard Gearhart uses AI as a vendor search engine, including to surface talent agencies for a CFO search, and warns that prompts and AI-generated reports have been treated as discoverable in court, including in a Delaware case where a manager’s ChatGPT strategizing was used against him.
  • Elizabeth Gearhart says AI image generation has improved so much that ChatGPT now produces studio-quality YouTube covers and can change a single word or color without redoing the whole image.
  • The panel splits on jobs: Cliff predicts hundreds of millions of eliminated jobs, deflation, and universal income, while James says workers who leverage AI will keep their jobs and Elizabeth thinks AI is creating jobs.
  • Eric says current language models turn language into math and miss emotion, and he sees an opportunity to rediscover our humanity. James notes that AI trained on human data still gets about a quarter of the engagement his own LinkedIn posts get.

Main Topics

1EVERYDAY AI

Hear how customer service, content, dashboards, and vendor search are being handled with AI.

2LEGAL RISK

Learn why prompts and AI outputs can be discoverable, and why public LLMs are not private.

3JOBS AND HUMANITY

Follow a debate over job loss, universal income, AI slop, and what makes us human.

Featured Guests

JT

Runs customer service and quality assurance with AI, keeping the human touch.

James Thornton
CN

Uses AI for decision support, content, and workflows, and expects massive job loss.

Cliff Nonnenmacher
EK

Uses AI prompts to turn locked infrastructure data into dashboards.

Eric Kanagy

Meet the Hosts

Elizabeth Gearhart
Elizabeth Gearhart, Ph.D.Chief Marketing Officer, Gearhart Law

Elizabeth Gearhart, Ph.D. is the host of Real AI Use Cases: Business Owners Roundtable, where she explores how businesses are actually putting AI to work. She is also CMO of Gearhart Law and co-host of the nationally syndicated Passage to Profit radio show and podcast.

Richard Gearhart
Richard Gearhart, Esq.Founding Partner, Gearhart Law

Richard Gearhart, Esq. is co-host of Real AI Use Cases: Business Owners Roundtable and founder of Gearhart Law, a full-service intellectual property law firm. A longtime podcaster, he hosts Intellectual Property News and is the host of the nationally syndicated Passage to Profit radio show and podcast.

The roundtable’s through-line: AI is moving fast, so use it with human judgment and expect the debate over jobs to continue.

Episode Summary

Frequently Asked Questions

How can AI run customer service and quality assurance?
James Thornton says his company’s entire customer service and quality assurance group is now powered by AI solutions, which makes it much faster to resolve customer concerns and to test quality. He says AI does not get you 100% of the way there in many cases, so human intuition and the human touch still matter.
How can AI help with decision support and content?
Cliff Nonnenmacher uses AI to pressure test ideas, for example working out that a 10% net margin business cutting its price by 5% must double in size to break even. He also says AI replaced a $5,000 a month content creator for about $20 a month, saving roughly $60,000 a year, and helps with voice, thumbnails, and distribution.
How can AI turn locked-away data into dashboards?
Eric Kanagy says customers in critical infrastructure often have data locked in old vendor interfaces and end up exporting it to Excel. With an AI prompt, they can generate dashboards in the columns and graphs they want, and a person who spent a week a month on water usage data can now generate the report in seconds.
Can your AI prompts be used as evidence?
Richard Gearhart says recent court cases have treated AI prompts and generated reports as discoverable, including a Delaware case where a manager who strategized with ChatGPT about breaking a contract was sued and his prompts were considered discoverable. He says public LLMs are like talking to the public, and an enterprise version may offer different arguments about confidentiality.
Will AI eliminate or create jobs?
Cliff Nonnenmacher predicts hundreds of millions of jobs eliminated, deflation, and universal income in many countries. James Thornton says leaders are split but would agree that employees who leverage AI will keep their jobs, and Elizabeth Gearhart thinks AI is creating jobs because humans ask creative questions AI would not think of.
What is AI slop and why does human writing still win?
Elizabeth Gearhart describes AI slop as taking whatever AI spits out and posting it. James Thornton says an AI agent trained to 99.9% of his voice still gets about a quarter of the engagement of posts he writes himself, because AI is trained on human data, and Eric Kanagy says current language models turn language into math and miss emotion.

Episode Transcript

View Full Transcript

This is Real AI Use Cases, Business Owners Roundtable, with hosts Elizabeth Gearhart, podcast consultant, marketing expert and PhD researcher using AI every day, and Richard Gearhart, entrepreneur, seasoned business owner and intellectual property attorney specializing in innovation. Here’s how real companies are using AI right now.

Now it is time for our next segment, AI Use Cases, Business Owners Roundtable. I know you guys are all up to your eyeballs in AI, but I’m going to ask you each for your favorite use and then we’ll talk about it a little. So James Thornton with DAZ 3D, what is one of your favorite uses of AI?

I would say that in our business, our entire customer service and our quality assurance group is now powered by AI solutions, and it’s pretty incredible how fast you can resolve customer concerns. We still often need to apply the human touch. I will tell you that I don’t think in many cases AI gets you a hundred percent of the way there, so there’s a lot of value to human intuition and human touch in that process. We’ve become so much more efficient at quality testing and assurance, and most importantly, outwardly facing, in how we are able to respond to customers and their needs.

Cliff Nonnenmacher with Franocity, what’s your favorite way you’re using AI in your business right now?

So I use it three different ways. I use it for decision support, to support decisions I’m about to make. As I said before, I believe in leveraging collective intelligence. There’s no greater intelligence to leverage than artificial intelligence right now, which is sad, because we’re all here on a podcast with human capital, and this is about to be worthless in T minus five years. Think about what I just said. I’m a podcaster, and everything that we’re doing, human capital will outlive its usefulness in probably less than five years. No one’s going to want to hear what we have to say, which is crazy. So I use it for decision support, to pressure test an idea. Does this make sense? Just test it and flesh it out and see, do the economics make sense? There was an interesting thing that I came across the other day. If a business owner had a 10% net margin and wanted to be competitive and cut their price by 5%, what impact would that have on their business? A 5% reduction in your product or service does not seem like a huge amount of money. But if you want to know the answer, it’s actually mind-blowing, so I’ll repeat it. If you had a 10% net margin and you cut your price by 5%, you literally need to double the size of your company to just maintain breakeven. Double. So I thought, that can’t be true, I have to flesh that out, and I just kept going through the math. So decision support, I use it heavily. It’s not my advisor on mental health, like some people use it as a mental health support system. It’s not my friend. I don’t use it that way. Decisions, content, and distribution. We used to pay someone $5,000 a month to create content for social media. For $20 a month we save 60 grand a year, just one person. So you think about just content, how to distribute that content. What does our voice sound like? Is it the right voice to have in the marketplace? What does each social media platform want to hear? How long, how much, what’s the message? What’s the thumbnail? Split test the thumbnails. If you ever listen to Mr. Beast, they will split test 50 thumbnails before the thumbnail hits the market. That has made him just a juggernaut in social media. No one has accomplished what he’s accomplished as a social media influencer, and I use it as operational leverage. We’re talking about workflows. Everything James said, my head was nodding. I’m in complete agreement with how you could use different ideas, different modalities, if you will, within this. We just say AI and it’s such a broad context. And then he says, well, I use AI for 3D caricatures or these avatars. That’s brilliant. I could turn them into cartoons, I could turn them into what looks like a human. It’s brilliant. So, CRM workflows, summarizing calls.

I disagree with you on that point, because AI gets it wrong a lot, and I play with it down in the weeds, so I’m not hiring somebody else to do the AI stuff for me. I’m doing it myself all the time, and I’m going to talk about an issue that I just recently had with it, but I do think you need human eyes on it. I really do.

If you’re having issues with AI now, you will not have those same issues, literally probably less than 12 hours from now. The rate at which it’s expanding, the rate at which it’s evolving, is just insane. We have not seen anything like this in a technological revolution ever in the history of the planet. Nothing has been this significant. It’s mind-blowing. We haven’t seen what’s coming.

Not since the invention of the wheel.

Eric Kanagy with SimpleSense.io, what’s your favorite way you’re using AI right now?

I work in a highly regulated space, so you can’t just try things out in that space. We work in critical infrastructure, security systems, fire alarm, building automation systems. The interesting use case we’re working on right now is that you have all of this data locked away in these arcane interfaces from these big vendors that have been around for decades. The end users want reports, they want to see that data in certain ways, they can’t get to it, and so they end up just exporting data into data files and moving it into Excel and trying to make it work. There are data visualization tools out there, but to build dashboards, you need some technical know-how. Well, now with an AI prompt, if you know how to write that prompt, which you do need to know how to write, you can generate those dashboards. So now there’s much less friction for that user to say, yeah, I want this data, I want it in these columns, I want to see this in this kind of graph. And it can very quickly iterate for that user so they can see the data exactly as they want it and get to that data. There’s a guy that we work with who spends a week a month generating water usage data. This is across a massive facility. He can now generate that report in seconds, something that used to take him a quarter of his time. That kind of savings is what we’re looking at.

So Richard Gearhart with Gearhart Law, what’s your favorite way you’ve been using it lately?

Lately, I’ve been using it to find vendors, using it as a search engine, and it tunes the search more specifically to what my needs are. Instead of just putting in a few keywords and then having a bunch of possibilities pop up, I can be specific about the characteristics that I want, and it’ll find vendors for me that I would never have considered before. So it’s a very straightforward, very simple way to use AI, but it is proving to be very valuable. We’re looking for a new CFO, and it surfaced some talent agencies that I was unfamiliar with and made some really great recommendations that are speeding the process considerably. So even though there are humans involved, it is doing things that weren’t available before.

Well, I’m Elizabeth Gearhart with Gear Media Studios, and I know you guys have probably all touched on this. A few months ago, a year ago even, I would try generating images in ChatGPT, and they were really terrible. So I’d go back to Canva, and then I found Manus, which is an agent, and it did better images. Well, then somebody said, oh, you should try ChatGPT again, and actually you should try Claude, which I haven’t gotten back to that deep yet. I started using ChatGPT for image generation again. I teach this adult school class, getting started in podcasting, and for the last class, everybody gets to make a little video in my studio. And so I took those videos and I did a screenshot of the people’s faces when they’re smiling, like I managed to catch every one of them. And then I put the screenshot of the person, the title of their podcast, and a little bit about it, because they didn’t even do a full podcast or anything, into ChatGPT. I said, okay, generate a YouTube cover for this. I was blown away. It wasn’t just that it could take the prompt and figure it out. It was the quality. I don’t think I could go into Canva and generate something like that.

That has changed so much over even the last three or four months. If you tried getting a thumbnail image out of AI three or four months ago, you just got this weird, distorted thing, and now you’ve got a thumbnail that looks great.

That’s amazing. But you know what else it can do now that it couldn’t do before? Before, it would do a thumbnail and you would say, okay, keep everything the same and change this one thing, and it would change the whole thumbnail. Now it can change one word, it can change one color. As Cliff said, the speed at which these things are going is like a steamroller. So I want to open the floor up. I do have another AI thing I want to talk about while we’re here, but I want to see what other people want to say about AI in general.

The attorney in me is going to come out and say there have been recent court cases where people have used AI for strategy purposes, and the prompts and the reports that were generated from the AI have been considered to be discoverable, and have actually ended up in written opinions of the court. There was one case in Delaware not too long ago where a senior manager decided that they wanted to break a contract with a company that they owned. He went to his legal team. The legal team said, no, you can’t do it. He went to ChatGPT and started strategizing with ChatGPT about how to break the contract, and then he followed ChatGPT’s advice, ousted the management team of the company that was the target, and then got sued for breach of contract. All of the prompts and all of the information that he had used in ChatGPT to create this program were considered discoverable information.

That could be used against you.

In the court of law, it could be used against you, and the court even cited some of the information in her opinion. So you have to be careful when you’re using this stuff. If you’re concerned about an employee, for example, and you’re brainstorming with ChatGPT about that employee, that could conceivably come up later in a dispute. Now, most prompts only have a limited lifespan in ChatGPT, but it depends on the engine that you’re using. And if you copy the output and you store it someplace, then it’s considered fair game, especially if the LLM is a public LLM. You may have some different arguments if you have an enterprise version and you can retain the confidentiality of the subject matter. But if you’re using a public LLM, it’s pretty much considered talking to the public about it. So I

I think if you’re using a public LLM, it’s like you’re running down the street naked. Everybody can see everything you put on there if they want to.

You have to assume that it could come back and be part of any

labor legal issue.

Yes.

I was listening to both Cliff and Eric and their comments, and they both said things that I think are really interesting, especially in this context. I think if you were to survey 500 executives at large companies, people that are leading AI for some big businesses, I’m finding it’s almost split down the middle whether those leaders think AI will create jobs or will eliminate jobs. Cliff was talking about the loss of human capital. What I think they would unanimously agree on, though, is that the employees that are using and leveraging AI in their workflows will be those that get the jobs and retain the jobs. So you just can’t put your head in the sand. And then Eric talked about this report that somebody would take a month building and they can do it in seconds. What I’m experiencing, and this might tip the direction of what I think, whether it creates jobs or eliminates jobs, I do still believe that right now, and even in the foreseeable future, there does need to be a combination of AI with human intelligence and human intuition. When I think about being able to generate reports that took months to build, you can generate those in seconds now. What that allows the person generating the report to do is spend less time building it and more time analyzing it, leveraging it, and doing their job better with it. And so, to me, and again, being a very AI-centric business, I find my employees feel more efficient. They feel more satisfied, but they still feel like they have a role in the process. You mentioned the hallucination that LLM models still do. There’s a viral clip out in the market today where the guy asks AI how many Es are in the word seventeen, and AI comes back and says three. And the guy says, no, there’s two. And it says, oh, you’re absolutely right, there’s two. And he says, no, again, there’s 20. And AI said, I can see why you would arrive at that conclusion, but I’ve recounted them and now I can tell you definitively there’s five.

James, to your earlier point, I think where humans are necessary is, and I’m a scientist, I have a scientific background, we think of the questions. It gives you a lot of follow-up questions, but when it comes to fundamental core questions and the prompts that we do, I think we ask creative questions that AI would not think of. AI to me is very linear in its thinking. But I think it’s creating jobs.

So I would like to hear Cliff’s response to your comment. I must be the contrarian here. She thinks that AI is going to take over everything in five years.

I think you’re living in peak inflation right now. I think AI, robotics, humanoids, and quantum computing are deflationary. I think the thousand-dollar iPhone in everyone’s pocket will be 200 bucks. It’s not going to cost peanuts to make it. I think there will be hundreds of millions of jobs eliminated across the entire planet, and I think you will see universal income introduced in many countries because of it. I don’t think people comprehend where this is going, and I don’t think people comprehend quantum computing and what it’s about to do. Just the discussion of quantum computing has affected the Bitcoin blockchain conversation and valuation. If you start to extrapolate, right now we’re talking about hallucinating. This thing is like six months old in the scheme of where it’s headed. It’s like being mad at a baby because it can’t pronounce a word. It hallucinates, it doesn’t ask the right question, it doesn’t prompt what this thing is going to do. It’s going to replace all of us. It’s insanity where we’re headed with this. If this thing is not regulated at a government level, which it won’t be, because the bad actors are not going to regulate it the same way. The state of Florida wanted to tell everyone you have to have a carry permit for a weapon, until the state of Florida figured out that every criminal in Florida doesn’t follow the law, and then the governor said, you know what, why not let everyone carry. It’s these types of things. The United States wants to regulate AI? Go ahead and regulate it. Regulate AI, regulate crypto, because China’s not going to do it. Other countries won’t do it. Russia won’t do it. So now you’re at a disadvantage. So it’s not going to be regulated. This genie is out of the bottle. You’re not unringing this bell. It is going to have mass implications across countless industries. Not to mention, the next time I’m on the show, you will have a humanoid. Probably James will not be James. The next time I’m on this show, I’m going to have to ask James and Eric, are you actually James and Eric? And Eric says, no, I’m in Seattle right now in a board meeting, this is my avatar filling in for me, that knows more about any question you could ask, including the pronunciation of my last name. It’s crazy. Peak inflation, deflationary, all values come down. All the complaining about I can’t buy a house goes away. The complaining that the average age of a homeowner in America is 40, that goes away. Why these kids can’t have babies, that will ultimately go away, because they’re not going to want them anyway. There’s a lot to unpack here.

Except, Cliff, according to what you’re saying, maybe houses are cheaper, but they still can’t afford them because they don’t have a job.

Because they don’t have a job, until you’re on universal income.

Okay. Eric, you were pretty neutral. What do you think?

I’m going to try to be a little more practical than what Cliff is saying. But I think what’s interesting about AI is it does call into question, well, what does make us human? And what does a human interaction, what is the value of that? And what do we get out of that? If we just have avatars everywhere acting for us, then why do we even need them? What is the point of this interaction right now? What are we getting out of it? For a listener, understanding that all of us are real is going to bring that into question. I think there’s also an interesting reaction post-COVID of people wanting to get together in person again. We were all remote, and isolated, and so now we want that human connection. You can see it in marketing content. People sending me AI-generated sales emails, you can kind of tell, and it’s not a good look when somebody sends you something like that. I had one the other day where it was this really long email, and when I first looked, I usually don’t read these, somehow it caught my eye and I was reading it, and I had to put it into an AI detector because I couldn’t believe that this guy had actually written this specifically for me. It turns out he had. So I called him up and we actually had a conversation, because he had taken the time and written this thing out. That did catch my attention. Maybe eventually you can’t tell anymore, and then we have to figure out what that means. There’s an opportunity here to rediscover our humanity.

I’m doing a presentation, a basic AI one, but during my presentation I wanted to bring up AI slop. You all know AI slop, right? Have you heard that term? That’s when somebody puts a question query into AI and takes whatever it spits out and posts it. And it was really funny, because AI knows what AI slop is now. I asked it, what is AI slop? It uses certain words, it uses three things in a row, and it has a certain length to its paragraphs and its lines. And to your points, now that it knows what AI slop looks like, it’s probably not going to do it anymore. Right? And so it’s going to try to make it more and more human instead of less sloppy.

I can say there still is such an interesting tug and pull with what Eric said. And again, I’m not trying to challenge Cliff, but I’ll give you a very simple example of AI slop. If my AI agent writes a LinkedIn post for me, it tells me that based on the way I’ve trained that agent, it can get to 99.9% of my actual voice. Now, when I write a post myself, when I actually write the post, I get four times the engagement. So how is that AI agent representing 99.9% of my voice and I get 25% of the engagement I get when I write it myself? And the answer is, guess what? AI is trained off of human data. That’s what it’s trained off of. Now, not to get too technical, because I know we’re not supposed to, but what Cliff is describing is what I would call artificial general intelligence, where the model actually doesn’t need training data but can reason on its own, is not a linear thinker like you’ve talked about, Elizabeth, but can really think like the human brain. And I think that’s a long way off, no matter what you’re hearing. And what I know about providing training data to train AI models to get to that general kind of intelligence, where AI is its own brain and doesn’t need the inputs, doesn’t need the training data, we’re a bit off from that.

I would agree with that. So where I am on the Cliff side is, what’s coming is going to keep advancing and it’s going to go quickly. The current language models turn language into math, and what they miss is emotion. As we grow up and we learn words, there’s emotion attached to words and meaning, and the current language models can’t get that, because they’re using math to map words to each other. So there’s a depth of meaning missing, and when you read something from AI, it feels flat, and that’s why. It’s missing that richness of language that we understand from learning a language growing up. It will keep evolving and changing. There’s something better than the current generation that’s coming.

I’ve heard a little bit about it, but I’m going to challenge you guys. I would like to get this group back together in six months or a year and talk about AI again and see where we all are on the AI spectrum. It is changing so fast, maybe in six months you guys want to get together as a panel again.

I’m glad to come back, because I’m going to be right.

Yeah. Cliff, if you’re right in six months, then my avatar will humbly apologize.

So either you or your avatars will be back in six months. Okay. I’m going to make a note. In six months we’ll get you guys back up.

We need a little wager here.

Oh, you want a wager?

We need a wager of some sort.

I think AGI will be here, but it’s going to be a weird transition, and there’s not going to be a day when we wake up and all of a sudden it’s here. It’s going to be that things get taken over and it gets baked into what’s normal and we just don’t really notice it. I have a robot vacuum that goes around and vacuums my floor. I don’t vacuum my floors anymore. It’s not great at it, but it vacuums every day. I don’t really think about sweeping the floors anymore. And so that just happens. That thing was like two hundred bucks.

So you think it’s like the frog getting slowly boiled, right? You put the frog in cold water and gradually turn up the temperature?

Yeah. I mean, it’s our technology, so we should be in control of it. It shouldn’t control us. So that’s the opportunity as an entrepreneur.

That’s why I said at the beginning, I have to constantly tell it that old, smart blonde ladies have a brain too, because it doesn’t think that we do, because we’re not the ones programming it generally. This has been Real AI Use Cases, Business Owners Roundtable.

You have been listening to Real AI Use Cases, Business Owners Roundtable. We hope you found this valuable. Join us again for more stories, because the future of business is driven by AI. This podcast was recorded at the iHeart Studios in Manhattan as part of the Passage to Profit Radio Show.

Keep exploring the show

Hear how more business owners are putting AI to work in their day-to-day.