Jeremy Utley: Author of The Human Advantage

Episode 917

As AI gets smarter, what actually makes us more valuable?
On this episode of The Kara Goldin Show, I sit down with Jeremy Utley, leading AI and innovation expert and author of the new book The Human Advantage: What AI Can't Do Without You. Jeremy has spent years teaching creativity and helping business leaders rethink how they solve problems, and now he's challenging how we think about working with AI.
We talk about why the people getting the most out of AI aren't necessarily the most technical, how asking better questions can lead to better ideas, and why human experience, judgment, and creativity matter more than ever. Jeremy also shares practical strategies like reverse prompting, thinking out loud with AI, and using these tools to make smarter business decisions.
We also explore why AI can sometimes make teams less creative, the mistakes entrepreneurs are making, and how to embrace new technology without losing what makes us human.
This is a great conversation about AI, innovation, entrepreneurship, and the skills that will matter most in the future. Listen here or anywhere you get your podcasts. 🎧

Resources from
this episode:

Transcript

Kara Goldin 0:00
I am unwilling to give up. That I will start over from scratch as many times as it takes to get where I want to be. I want to be. You just want to make sure you will get knocked down, but just make sure you don’t get knocked out. Knocked out. So your only choice should be go focus on what you can control. Control. Control. Hi everyone, and welcome to the Kara Goldin Show. Join me each week for inspiring conversations with some of the world’s greatest leaders. We’ll talk with founders, entrepreneurs, CEOs, and really some of the most interesting people of our time. Can’t wait to get started. Let’s go. Let’s go. Hi everyone! Welcome back to the Kara Goldin Show. AI is changing just about everything, but here’s a question I think more people should be asking: As AI gets smarter, what actually makes us more valuable? So today’s guest is Jeremy Utley, one of the leading voices on AI creativity and human potential, he’s the author of the brand new book *The Human Advantage*, which I got an early copy of that just launched. So so good, *The Human Advantage: What AI Can’t Do Without You*. And he’s also the co-author of another incredible book. We had him on a few years back called Idea Flow, which if you haven’t read that book too, that is absolutely amazing. Jeremy has spent years teaching innovation and advising Fortune 500 leaders on creativity and growth, and his work has reached millions through Stanford, Harvard, and his podcast Beyond the prompt, but when Chat GPT arrived, even Jeremy realized he had to learn a lot about how humans and AI could work together. So his big idea: the people who get the most out of AI aren’t necessarily the most technical. And so many, so many thoughts in his in his book about how to actually grow with AI, as I like to say, instead of actually trying to block it. So so so good to see you, Jeremy. Very excited to have you here.

Jeremy Utley 2:13
Thank you. Thanks for having me. And I love that phrase, grow with AI. I’m totally going to steal it.

Kara Goldin 2:18
I you absolutely can and should and and it’s it’s definitely I I feel like it’s something that everybody thinks about. You and I were talking. I think so many people, you know, it’s a buzzword to so many people versus actually, how do I actually work with this in in my life? But your new book is called The Human Advantage: What AI can’t do without you. What’s the big idea behind it, and why did you decide to write it now?

Jeremy Utley 2:48
Well, I’ve had the privilege of working with now 10s of 1000s of professionals over the last four years, all of them grappling with how do I use AI, and what I learned is, in short the people who use AI underperform, and the problem is not AI. The problem is the word use. Most people treat AI like a tool, and they get mediocre outcomes. The people who outperform, the folks who I’ve studied who are outperforming, aren’t people who are opting out. There are people who are working differently with AI, and so they don’t use it like a tool. They treat AI like a teammate, and those people and that insight really fueled a renaissance in my own thinking. And it turns out it’s not technical gurus, it’s not rocket scientists, it’s normal humans who’ve learned how to get great work out of this weird intelligence called people. If they leverage those same skills, they can get great work out of this weird intelligence called AI. And there’s no technological barrier. It’s really more of a human ability that needs to be unleashed. And and that to me, treating AI proficiency more like a skill is something that not a lot of people were talking about.

Kara Goldin 4:05
That totally makes sense. And you’ve spent years teaching creativity and innovation, and frankly, bringing the best out of people who maybe are not the most technical or who are

Jeremy Utley 4:19
trying.

Kara Goldin 4:20
Right? I mean, you you have you are like the master of this. So so much conversation about AI replacing jobs today. Do you think we’re asking the wrong questions of organizations versus should we be thinking more about how AI changes the value of what the existing humans bring to work, bring to an organization, bring to life.

Jeremy Utley 4:45
Yeah, I love what Jensen Huang said. Jim Cramer was interviewing him recently, and he said, “If AI is so amazing, what’s up with all these layoffs?” And I loved what Jensen said. I don’t know if you heard the conversation. He said, “These companies have no imagination.” He said, “We have.”. At Nvidia, we’ve got a list of dream projects a mile long we’ve never been able to get to, and finally, because our people are augmented and unleashed by AI, we’re able to do stuff we’ve never been able to do before. But that is not true of many organizations. Most organizations, the honest truth is, they don’t have a dream list of projects. The only thing they can think to do is the thing they already do. And if the only thing you can think of doing is doing what you already do faster, then there are very few options other than I guess we don’t need as many people now. But for folks who see that the primary limitation is their imagination and dare I say their ambition, and if they would instead of ratcheting down their head count, they would ratchet up the level of ambition for the organization. They would see possibilities with AI that are completely. It’s not about 10% faster, 10% better. It’s about what’s 10x. What can we do now we’ve never even dreamed of doing before.

Kara Goldin 6:01
So two or three years ago, you really started thinking more and more about this. What did you believe about AI two or three years ago that maybe you’ve completely changed your mind about

Jeremy Utley 6:15
it? Oh my goodness! I mean, there. How how much time do we have? There are so many things, right? I mean, starting with it’s like it was it was intimidating, right? It’s deeply technical. I’m not technical. I said finance. I got a business school degree, but then the other thing I would say is even from the very beginning, Kara, I felt like I was behind, and actually, a number of people that I’ve met feel behind, and they’re self conscious of the fact that they’re, you know, everybody’s hurtling forward faster, and you know, I had interestingly, I I have a course on my website. Anybody can take it. It’s a three week long AI boot camp. I think it’s brilliant. I built it for grad students who are asking me how to get started. So I built this course, and then I made it available to anybody. And like any good email marketer, if if you don’t sign up for the course, I ask you why, right? Is it is it too expensive? Is there not enough time? Are you interested in AI? I had a Stanford professor reply to this email. He’ll remain nameless, but the Stanford professor wrote me back, and here’s what he said: direct quote. And the choice that the question is, why didn’t you enroll in this wonderful kind of AI proficiency course? And the answers are four. One, too expensive. Two, I don’t have time. Three, I’m not interested in AI. Four, it’s unclear what I’m going to get out of the course. And the Stanford professor wrote me. He said, “Dear Jeremy, fear of learning something new that I should effing already know wasn’t one of the choices.” And that was I thought it was so brilliant because it encapsulates this sense of I know I should know more and I don’t, so I guess I’ll just bury my head in the sand.

Kara Goldin 7:46
Yeah,

Jeremy Utley 7:47
and I think for me, even I mean, I remember Christmas break of 2022. Okay, ChatGPT is a month old, and I started enrolling in a class, and I actually thought, I bet it’s too late. I actually had that thought. I shouldn’t waste my time. It’s too late. And then a year later, I see another company is rolling out an AI kind of assistant programming program, and I go, Nah, I should do it, but it’s too late. And I would say I wasted far too much time feeling self-conscious of the fact that I was behind. And here’s what I want to say to everybody, maybe in your audience who may feel that way. Andre Karpathy is probably the greatest programmer in the world, or you know, he’s top 1% for sure, founder of OpenAI. He tweeted the other day, “I’ve never felt more behind as a programmer. If Andre Karpathy has never felt more behind, it’s perfectly normal to feel behind. In fact, I would say, if you don’t feel behind, you’re delusional. The question is, given that you feel behind, what do you do? And that is where I try to help people take very simple, pragmatic steps towards proficiency and towards mastery. Instead of being overwhelmed and paralyzed, just move forward with confidence. That that to me is is something that everybody needs a little bit of help and structure with.

Kara Goldin 9:09
So one of the things I loved in your book, The Human Advantage, was the topic of the prompt, the perfect prompt. What do you think people have wrong? And it’s really retraining your brain, which again you have years of experience on on this topic. But what is the first step that people need to think about when they’re getting that perfect prompt?

Jeremy Utley 9:31
I mean, I would stop thinking in terms like the the the the entire problem I have with even your question is it singular prompt singular right?

Kara Goldin 9:40
Yes, yes, yes. And

Jeremy Utley 9:41
what really what really counts is actually a conversation. It’s a dialog. It’s a back and forth. You know, most people, if you look at their chat history, there’s no chat. It’s a search query. It’s a single turn question answer done. Right. There’s no chat. There’s no conversation. Right. And the thing that’s amazing about these conversational engines is you can go, you can give a response, you can give feedback, you can push back, you can ask for a different perspective, right? So I I think Anthropic did one of the biggest studies of kind of what what they consider to be proficiency and getting better outcomes, and the single biggest variable they found that determined the success of a user’s experience was number of conversational turns, and that’s something that very few people. I remember I was talking with the with the head of Harvard Simulation Lab in the medical school, and I asked her, “Hey, what are you using ChatGPT for?” And she goes, “Oh, it’s great. I give it a grant proposal, or I give it a grant kind of a you know form, and I ask it to craft a draft of my proposal, and then I get the draft and I put it in Word and I edit it and all that stuff. And I said, “Doctor, do you tell Chat how you edited the proposal it gave you?” And she, I actually thought the Zoom had froze for a second because she got so still, and she goes, “Honestly, I never thought to talk back, and I think that just the paradigm of this is a conversation. You know, like one of the early one of this, we did an early study myself and a research partner at Singularity University, wonderful guy, Kian Gohar. We did a research study that ended up getting published in HBR, a bunch of different places, and one of the key quotes was the CIO of a company that we studied, read our report and goes, “Oh, you mean I should put the emphasis on the chat, not the bot?” We talk about chat bots. I think in terms of bot, but you’re saying chat, and we go, “Yeah, it’s it’s a conversation.” As we said, like one of one of our ways of describing it was more chat, less bot. And I think for anybody, if they would think more in terms of bringing their own humanity to the interaction, I mean, the great irony right now is computers are finally able to understand us, and guess what? We’re the ones who sound like robots.

Kara Goldin 11:54
I know, which is just ironic.

Jeremy Utley 11:56
And if we would be more human in our interaction, it would. It turns out the outputs of an AI model are a function of the inputs to the AI model, and everybody thinks a lot about what’s the training data, whatever. But very few people think about how am I showing up as a collaborator, and it turns out that the single biggest point of leverage you have as a human is the way you’re showing up as a collaborator, and the people who who think about that and who are thoughtful about it and careful about it, they end up outperforming with AI.

Kara Goldin 12:29
So reverse prompting and how that can help founders, or how could founders use that when considering a new launch for either a business or products, or should I head in that direction? All of that.

Jeremy Utley 12:47
Such a simple technique. So what Kara mentioned, it may sound technical. It’s not. Basically, reverse prompting is giving the model permission to ask you questions. You know, most people start with the assumption: I, the human, am the question asker and the AI is the answer giver, and what the research suggests is that’s a faulty assumption. It turns out AIs are incredibly insightful question askers, provided they’ve been given permission to ask questions. So, as Kara mentioned, reverse prompting is a technique where just put at the end of any prompt or any question or any task you’ve given an AI. Just tell it the following: some version of the following: Before you respond to my query, please ask me questions until you have enough information to deliver an exceptional result. The model then thinks my job is not answering this question. My job, which every model wants to be a helpful assistant, but the question is what what is helpfulness. And if you just say write me this email, then the then the AI goes helpfulness is writing the email. But if I say instead of just write this email, I say using the prompt that we just discussed, I say write this email. But before you do, please ask me any question you need to do a great job. It defines helpful as asking good questions to be able to write a good email, which is very different than write an email, right? And so the the model then asks you great questions. You end up giving it much more information, which enables it then to do a better job of delivering the email, all because if just like any good manager, what’d you say? If anything’s unclear, feel free to ask me any questions. My door is always open, right? We’d say that to any capable teammate, a human teammate. Has have you ever said that to AI?

Kara Goldin 14:34
Yes, so so interesting. Your research found that that many teams are using AI and becoming more confident, I also see that many companies’ teams are doing enclosed units of AI within, you know, their company. They don’t allow many are not allowing, at least while they’re working, to use, you know, an open system. Do you have any? Thoughts on that? Do you think that that is? I mean, I know the reality is you can go home and you can grab your computer and or your phone and and do it. But what what are your thought? Do you think that companies who are doing that are still not really understanding the way this all works? Or

Jeremy Utley 15:17
I I think they’re doing the best that they can. I and I sympathize greatly with IT leaders and CIOs who are trying to keep things secure and compliant and all that stuff. I work with, for example, nuclear contractors who are under very strict DOE guidelines, and they’ve got to bring two machines to every class. You know, I I recently did a talk at Stanford with the CEO of a nuclear contractor saving $5 billion on a major nuclear project because of the augmentation that AI has brought his team, and he’s doing it in an incredibly limited environment. So, the starting point, the first thing I’d say is, don’t allow your technology limitations to limit your ambition or imagination. I mean, as they say, necessity is the mother of invention, right? And you can take whatever limitations you’ve been given and do far more with them than you think. I mean, I would I would argue, even somebody who has the most limited system, I have not seen somebody who’s who’s deriving more than 10% of the total possible value they could be from their current setup. Most people are deriving maybe 1%. So if you think about like there’s kind of two concentric circles, Kara. One circle is what are the capabilities available to me, and a lot of people go, “I wish I had more.” Right, but the circle inside of that is what percentage of the capabilities available to me am I currently making use of? And for almost everyone, that’s a circle they can they can the stuff that you’re currently taking advantage of can grow a lot. Before then, you need to make the case to IT to grow my toolset. So that’s the first thing I want to say is I want to challenge every person who’s blaming compliance for not being able to do something. I think you’ve got an imagination problem, my friend. That being said, okay, so that’s kind of the first thing is all of us need to take personal responsibility for kind of maximizing our effectiveness, and actually, I would say one thing else there. Maximize, I think, is the wrong word. That’s that’s abstract. It’s kind of intimidating. I would actually say instead of thinking how do I get maximum use, it’s how do I get more value just relative to yesterday. How do I get more value today and tomorrow? How do I get more value from AI than I do today. That’s a far more practical question than maximum. But anyway, that being said, about because I see a lot of IT leaders and CIOs getting thrown under the bus. You know, compliance won’t let me do it. I rarely think compliance is actually the limitation for most people. So, that being said, I do think ultimately some teams and some organizations will limit themselves through the systems that they are architecting. Generally speaking, for the purpose of safety or you know some other purpose, systems integration, etc. And there’s there’s this concept called the bitter lesson, which is the following: AI systems can be trained for a very specific narrow task, say playing chess, or AI systems can be trained for very general purpose like language. The bitter lesson is, as you scale up language models, they outperform specific narrow models. So a language model like GPT Astra can play chess better than the the best world class chess model could a year ago. I mean, and many investment banks have learned this the hard way, right? Like, I think I don’t know if it was BlackRock or there’s one big bank you can bleep that out if I was wrong. There’s one big bank that spent like $10 million, maybe it’s Bloomberg training a proprietary financial services AI, and when the next model of GPT, which is at least two models old, when the next level of GPT came out on benchmarks, the general purpose GPT outperformed Bloomberg’s custom based model on every single benchmark of financial performance. Wow!

Jeremy Utley 19:02
And so, all that say, I think sometimes if organizations are too narrow and they go, “Oh, we’ve got this very niche, bespoke data set that we want to curate towards the lesson that is now known as the bitter lesson,” which has been learned over and over and over again, is scaling up general purpose models almost always outperforms, and so if you limit yourself and your people to hyper specific models, maybe they’re best today. In a year, they’re not going to be.

Kara Goldin 19:32
So interesting. So the biggest mistake you see entrepreneurs making today, because I feel like there’s so many choices out there with you know whether it’s you know OpenAI or or Gemini or and it I feel like people there’s almost too much choice right now and people are trying to figure out exactly what is the best what is the best thing to use what is what is the problem that they have. I think is another big thing because people don’t even know what they should be using AI to do. What if you were starting a company today with limited resources? Where would you use AI first, and and who is like the star out there for that specific example?

Jeremy Utley 20:21
The the the best thing that I would suggest is every single person, every single function should be AI native, operating in an AI native manner every single day. Every person, every function, every day, and the mistake is to place value, disproportionate value, on handcrafted artisanal cognitive labor. Right? You would not if if a cardiologist prided herself on handcrafted artisanal cardiology, and you go and you say, “Am I gonna get an MRI. She goes, no, no, no, no, no. I just use my ears. I practice handcrafted artisanal cardiology. You’d go, I’m gonna need a real doctor, right? Because a real doctor uses every means of delivering the best possible care. I think every single function in a business: finance, HR, marketing, product development, engineering, partnerships, legal. Every single function has the potential to be augmented. And so, if I’m starting a business, what I want to make sure is every single person is AI native. They are finding the most AI native way of accomplishing Their function, and this it looks different. But I’ll give you one example. I had a great founder on my podcast, my AI focused podcast, the other day, and they’re building tools for teams. He said they have a rule in the company: any time you have to open a browser, write an engineering ticket Because having to work through a web browser is evidence that their product is deficient in some way. They believe an AI-native company nobody should be working in a browser for any reason. And this guy told me he’s raised $75 million just this year. He hasn’t opened his email this year.

Speaker 2 22:21
Wow!

Jeremy Utley 22:22
He hasn’t opened his email. He hasn’t composed or read a single email in his email because everything goes through Claude code, everything. And so, to me, it’s just it’s a radical recalibration of what does it mean to be AI native in every part of my business. I think teams that really attend to that, you know, every single function defining a Stanford proficiency for themselves. Those are going to be the teams that are lapping kind of pre AI businesses in the near future.

Kara Goldin 22:52
You have a great story in the book about the Portland Trail Blazers. Can you talk about that? I thought it was a lot of fun.

Jeremy Utley 22:58
Oh my goodness! Yeah, I mean, I’m I’m a huge fan of what Dwayne, who’s the president there, and and David Long and my friend Christa Stout, many amazing folks there. What they did, David Long and Krista kind of took the the banner of AI and said, “How are we going to unleash the organization?” And they ended up hosting. They they started by building a bunch of tools for other people. They thought would be helpful, and shocker, not a lot of people took them up. And then what they did is they said, “Hey, on we need to take a more human-centered, almost design thinking kind of approach.” And they hosted what they called luncheon launches, and they didn’t mention AI at all. They just said very simply, “Fill in the blank. In my job, it sucks that blank, and they just have people fill in the blanks. What sucks? And I want to, on the one hand, apologize for my French there because I like I’m from you know I’m from the Midwest, right? I hear sucks. I’m like ooh, right? But that’s actually the point. It’s not. It’s moderately annoying that right. I don’t want to hear about your moderate irritations, what sucks, and then what they would do is from these lunches they would distill pain points as opportunities for new AI powered solutions, and then they would build them in partnership with people addressing the thing that really bothered them. And what they found is they got much more buy in, much more engagement, and they were able to transform functions that that where where things were previously assumed to be given, they were able to reinvent from first principles by taking that kind of bottom up approach.

Kara Goldin 24:35
That’s so so interesting. Again, the book is The Human Advantage. Everyone’s got to get their hands on this. I felt like it was really uplifting as well. Like everyone’s kind of,

Jeremy Utley 24:46
I I hope so. Yeah,

Kara Goldin 24:47
like

Jeremy Utley 24:48
I hope so. You know,

Kara Goldin 24:48
everybody’s like, oh, AI. You know, this is haven’t people seen the movie Wally? And you know, it’s just it’s.

Jeremy Utley 24:56
There’s honestly, there’s never there’s never been a better time to be alive, Kara. There’s. If you want to create value in the world, there’s never been a better time to be alive, and that’s what I would say to everybody. The reason it’s so uplifting is because I’m super excited. I’m really optimistic. This is the this is an incredible time to be alive. I’m excited for my kids. It’s an incredible time to be alive. One thing I know you want to point people to the book. I will say, let’s make. I know that this podcast is going to come out like the day after the book. I’m happy to make all the pre-order bonuses available to your listeners through the end of this week. Amazing. So I’ve got like I’ve got a video course. I’ve got I’ve got all of the exercises in the book or in a PDF. I’ve got a new AI platform called the Advantage Engine, which basically interviews you about your job and then gives you customized recommendations for every single chapter and every single story for you in your life. I’ll give that to anybody who orders the book this week through here. We’ll just set up a link in your show notes or whatever.

Kara Goldin 25:53
Okay, perfect.

Jeremy Utley 25:54
Even though it’s not technically a pre-order, is what I’m saying.

Kara Goldin 25:56
Got it. That’s that’s terrific. And for someone, I just want to leave people with this for someone listening who is like, I got to get going. I’ve heard this. I’ve heard Jeremy talk about it. I’m going to buy the book. But what’s the one experiment they can try today to really feel like, okay, I got started?

Jeremy Utley 26:17
Okay, do this. Open an AI of your choice can be Gemini, Claude, ChatGPT, even Copilot. Doesn’t matter. Tell it the following: You are an AI expert. You’re gifted at helping professionals find non-obvious and valuable use cases for AI in their work. Please interview me, asking me about my KPIs, my workflows, my irritations, and my goals, until you can make at least three non-obvious recommendations for things I should be doing with AI right now.

Kara Goldin 26:53
I love that. Well, Jeremy, thank you so much for joining me today. I love seeing you again and the book.

Jeremy Utley 27:01
Likewise, the

Kara Goldin 27:01
human advantage. What AI can’t do without is terrific. I know it’s going to do so well. Jeremy is also on the speaking circuit. He also does workshops with companies. So definitely check out his website, Jeremy utley.com. You have so much that we can learn from from you. I mean, you’re you’re just. I’m so inspired every time I think about everything that you’re doing and and read. You write some amazing things on LinkedIn, Substack as well. So really, really great. So until next time, thanks everyone. Thanks again, Jeremy.

Jeremy Utley 27:39
Thank you.

Kara Goldin 27:39
Thanks again for listening to the Kara Goldin Show. If you would, please give us a review and feel free to share this podcast with others who would benefit. And of course, feel free to subscribe so you don’t miss a single episode of our podcast. Just a reminder that I can be found on all platforms at Kara Goldin. I would love to hear from you too, so feel free to DM me. And if you want to hear more about my journey, I hope you will have a listen or pick up a copy of my Wall Street Journal best-selling book, Undaunted, where I share more about my journey, including founding and building. Hint: We are here every Monday, Wednesday, and Friday. Thanks for listening, and goodbye for now.

Click to subscribe:

Listen on Apple Podcasts
Listen OnApple Podcasts
Listen on Spotify
Listen OnSpotify
Listen on Overcast
Listen OnOvercast
Listen on PlayerFM
Listen OnPlayerFM

People Also Liked

915

Krishna Kaliannan: Founder of Catalina Crunch

914

Lulu Ge: Founder & CEO of Elix

913

Dr. Dennis Gross: Co-Founder & Chief Science Officer of Dr. Dennis Gross Skincare

912

Sophia Edelstein: Co-Founder & CEO of Pair Eyewear

911

Saif Khawaja: Co-Founder & CEO of Seremoni and Shinkei Systems