Why Cutting Costs With AI Will Put You Out of Business with Steve Brown

Full Episode Transcript

This is the full recorded conversation from The Human Protocol Podcast, published by The Human Protocol Group. © 2026 The Human Protocol Group. Please link to and attribute the original episode when referencing this conversation.

THE HUMAN PROTOCOL (00:02.291)
Every company is racing to use AI to cut cast, and most of them are about to cost the one thing they cannot get back. Their people, their culture, the reason anyone chose them in the first place. My guest says there is a fork in the road, and if you take the wrong turn, you end up competing on nothing but price. Stay with me. Welcome to the Human Protocol, the podcast about staying human while building.

The Future with AI. I am your host, Michael Salman. My guest today spends his year in boatrooms telling leaders a truth they don't always want to hear. Steve Brown is an AI strategist, a keynote speaker, 55 talks across three continents just last year, and the author of a new book called The AI Ultimatum. His message: AI is not an IT upgrade.

And it's not a cost cutting tool. It is a business transformation. And the leaders who treat this as a chance to amplify their people will pull away from the ones who use it quietly to hollow themselves out. Steve, welcome to the HEMA protocol, my friend.

Steve Brown (01:16.246)
Michael, it's great to be here. Nice to spend some time together.

THE HUMAN PROTOCOL (01:20.103)
Yes, I appreciate you for taking some time to be with me and having this this conversation. I know this is a topic it is that you're passionate about as I am, so I'm very much looking forward to this. And and to get started, Steve, I want to ask you if you can share with us part of what you see from the stage that most AI commentary and news missing today.

Steve Brown (01:50.702)
I think the biggest

misnomer, misapprehension, misunderstanding amongst leaders about AI is that it's just another technology out of the IT department. Just like anything else that's come out of the IT department for the last 50 years, it is there to reduce costs and improve efficiency. And that's a lens that they have applied for 50 years and it's done done them well. They've got to their, you know, high positions by doing that well.

And they're treating AI like it's just another technology, and it is not. And that's the big mistake that I see leaders making. They're not recognising that this one is different. And the difference is that this can be used to amplify your people, not cost reduced r reduce and replace your people. So this is about labour amplification, not labour substitution.

And those leaders that go down that labor substitution path are going to destroy value in their companies and ultimately put themselves out of business. And that's my biggest fear. if if I have a personal mission, it is to save as many of the cherished brands that I care about as I possibly can. And so I'm spending my time with boards and leadership teams helping them see this.

and understand that the way to create value and to collapse costs and time and deliver better value to people is to use AI in a very different way.

THE HUMAN PROTOCOL (03:28.418)
Yeah, you know, I'm coming from the customer experience side of things, right? spent all my life working with delivering value to customers. And part of all this conversation and part of your focus and and what you worry about is when people connect to a brand, there is a specific things they come to this brand and they stick to it, right? And the last thing I also want

is that AI makes all the brand look like themselves, no difference between each other, kind of nothing that differentiate them from each other. And and and that that would be something that

We're not looking forward, right? We still want the brands to feel authentic, that there is a purpose and a mission and what they do, and to use AI in a way to back to your point, that is smart and that it continues to amplify what differentiates them from the others.

Steve Brown (04:29.719)
Yeah, how do you use AI to accelerate your ability to execute on your corporate mission, to deliver on your corporate purpose? how do you use AI to amplify your ability to keep your brand promise? That's what it really comes down to. And my worry is if you use AI for labor substitution and you just try and do some cost cutting, what you do is you might get some short to medium term.

Benefit for the shareholders. You may see the ability to bump your results by 10 or 20%, maybe 30%, if you do really well. But if you follow a different path, if you use AI to amplify the impact that you can have, what you're doing, you potentially if you pair human intelligence with machine intelligence and you do it elegantly, you do it thoughtfully, you could amplify your ability to deliver value.

By 10x or 20x or 30x. You're talking about exponential growth. And in so doing, you know, you get economies of scale, you're going to collapse your costs anyway. So that is the right approach. You think about two different companies that follow those two different paths. Company A goes down the traditional path. They maybe do incremental growth, they get, you know, 10, 20%. Now think about another company.

B that goes down that same path. If those two companies have equal access to capital and technology and they cost-cut their way and get as rid of as many people as they can, what have they created? Two companies that look exactly the same and all they can do is compete and to the bottom on price. And that's not good for anybody. Now think about company C that goes down that other pathway.

They amplify their ability to deliver against their core purpose. They maybe find other ways to deliver against their purpose they've never never done before, taking them into new businesses. And they have that extreme growth because they really embrace and lean into agents, if there's a physical component to what they do, robots, to amplify and scale their operations. Scale your operations, costs collapse. Now have those two three companies compete. Who's going to win?

Steve Brown (06:52.555)
It's gonna be the company with the scale and the reach and perhaps even going into new businesses and redefining what the game is and delivering new services in new ways. They're the ones they're gonna win and companies A and B fade through relevance. And that's my big worry, is that leaders are going down the wrong path and they don't that they they won't see that until it's too late.

THE HUMAN PROTOCOL (07:17.035)
And why do you think is that Steve? And what is the is the hyper focus on just the focus on getting efficiencies out out of the implementation of AI and not trying to focus on long term what this means for our

Steve Brown (07:34.254)
I think if you if you've been to the Super Bowl and won thirty times in a row with the same playbook and then the game changes, the rules change, if you're d you're tempted to play well, we won thirty times, why wouldn't we just go with the same playbook? And so I don't blame leaders, right? This is all they've known, that's what they grew up on, it's what made them successful, it what's what made their shareholders successful and happy.

But the game has changed. And this is a unique time, a unique technology or set of technologies. And leaders have to recognize that. That the game has changed. And and the goal has changed, right? You you're now if you cost cut your way you can't cost cut your way through this transformation, right? You have to embrace it because up until now.

You're a good manager if you're able to manage scarcity. This is the fundamental difference. You're managing scarcity, which means you have limited resources, you have limited headcount, you have limited budget. I've been a manager and leader for many years. This is what you're good at, right? How do you get the biggest results by managing with the headcount you've got, the capital equipment, you know, and facilities you've got, and the budget you've got. And every year you are Yeah, exactly. Yeah.

THE HUMAN PROTOCOL (08:48.852)
Less.

THE HUMAN PROTOCOL (08:55.122)
Deliver more with less, right? That all that all said.

Steve Brown (08:59.647)
Exactly. And that is living in a world of scarcity. AI offers us and affords us this world of abundance. I know it's a bit of a foo foo la la term to say this, but it's re it's real. It enables you to scale your operations at quite low cost. And I know people are complaining about the cost of agentic AI and AI in general, if you want a leading edge model right now. But just wait six months. You know, the cost of these things is dropping too much.

10x or 100x year on year for sort of a fixed unit of intelligence. So the type of capabilities that you get with a leading edge frontier model today will be available perhaps on your desktop computer, embedded in it, you know, a year from now. So the cost of delivering intelligence at scale, your ability to expand your reach, to come up with new products and services, to go to market in new ways, to create value in new ways.

just blew up exponentially. So you now live in this world of abundance. And as a leader, your job is to rec one, recognize that, to shape your level of ambition accordingly. And I think the best leaders will be the ones with the best vision of what they really want to achieve, which means you have to understand the possibilities this all presents. And then are you able to inspire your teams and and sh shepherd them in a way where you can

Walk into this world of abundance. And you do that by building teams which are a combination of human talent and machine talent. And your job as a leader now is to make sure that they can work together optimally to deliver against the vision that you set. That's it. That's the new game.

THE HUMAN PROTOCOL (10:46.116)
It's it's the new game, it's humans plus machines and how we can move together. Really I'd say one thing, what w what it used to be just human beings that now this new future it's AI agents plus humans or humans plus AI agents. And it's it's a def different set of rules that this new future is pushing us to start thinking about, right? And not only how to manage the human cap capital.

But also all that part of the infrastructure that comes in.

Steve Brown (11:17.035)
It's a completely different mindset. That's the biggest thing. It's it's thinking about your role as a leader differently. Right. You have to show up differently. You have to care about different things. You have to focus on different things in order to ultimately do the right thing for your shareholders, for your brand, for your employees, for your customers.

THE HUMAN PROTOCOL (11:19.417)
Yeah.

THE HUMAN PROTOCOL (11:41.113)
Yeah, it's especially when the roles and the job as we know it today, we know it's gonna change, right? There's a lot more that we on the leadership side we really need to think if we want to build this organization for the future, there are gonna be hard conversations about meaning and purpose. What is it that I have been doing for the past five and ten years that are not gonna be part of my job anymore, right?

And we're gonna be judging and making decisions probably on things that AI is doing. And the responsibility may continue without

Steve Brown (12:18.849)
Yeah, and and you know, if you're creating if your job as a leader and manager is to build high functioning teams of humans and machines working together, that sort of leads you to the question, how do I divide the work? What should the machines do and what should the people do? And you you know, you're looking at their strengths and weaknesses. What what we're doing is we're moving it through a sort of phase thing where we move from humans doing the work to machines doing the work and humans now overseeing

And guiding the work. So the machines do the work and the humans decide what work is valuable, what work sh is worth doing, is worthwhile. And that's where the purpose comes in. It's where human judgment comes in, human taste, right? How should the work be done? What is it that we're trying to deliver and in what way? all of that is very human work and is not going away.

THE HUMAN PROTOCOL (13:19.566)
Especially the conversation around what good looks like, right? And to get started, what is it that we as human and what is it as part of our job, what is it that we want this to remain human? And what is it that we are we are gonna be handed it over to to an automation or or AI to handle? And those are the type of conversation that should continue to drive where we where we go next.

Steve Brown (13:25.109)
Mm-hmm. Yeah.

THE HUMAN PROTOCOL (13:49.73)
But from what I know from you, you are an optimistic, right? And and that's part of the reason why you are having this conversation. Really? So that's part of the reason why you're having this conversation to bring visibility to leaders. Really, these are the things that you need to start thinking about. because if not, the future will look much different than what you are anticipating to look.

Steve Brown (13:55.262)
I like to think a realistic object. Yeah.

Steve Brown (14:06.221)
Mm-hmm.

Steve Brown (14:18.037)
Yeah, I I mean I call myself a futurist. I have been a futurist at Intel. I worked in Intel Labs ten plus years ago. And more recently I was an in-house futurist at Google's AI labs in London, DeepMind. so I call myself a futurist. That's that's what I that's how I show up for my clients. But I tell my clients, I deputize you all as futurists. Everybody listening to this, I deputize you as a futurist. And a futurist

It's not like a fortune teller, you know, where you try and predict the future. That's a that's a fortune teller. a good futurist Well, yeah. people think I like go into a dark room and smoke drugs and come up with big thoughts. That's not that's not what it is, right? a good futurist asks and answers two questions or helps to answer two questions. And I encourage you all

THE HUMAN PROTOCOL (14:48.684)
Thanks for letting us know.

THE HUMAN PROTOCOL (15:00.588)
Ha ha ha

Steve Brown (15:11.821)
To engage in this, right? We all have agency in deciding what the future is that we want to collectively build together. So the two questions are: what's the future that we want to build? It's the optimistic part. And what's the future that we want to avoid? Here you have to go on to the dark side and really think about what do we want to not build. perhaps we should have done a bit more of that when we were building the social media infrastructure that we have today. but you know that.

Those are the two questions and I I encourage everybody to engage in that. Talk about those two big questions with your friends, your business colleagues. if you're involved with people in government, talk to them about it. Because we need to shape this. It's going so fast. It's going so fast. And we all need to participate in helping to decide what the future is that we want to build collectively together.

THE HUMAN PROTOCOL (16:03.285)
Yeah, I totally agree with you. And let me tell you something from personal side. I absorb news coming from both ends. I think part of how we you know receive news and try to determine what's good for us or not, we need to try to understand where things may be heading to your point, what is the positive angle and what is the negative.

The the the the issues that I'm seeing today, Steve, is as you were saying, this is going too fast, right? And everything that goes fast without us having a say on it, then we kind of lose you know the ability to try to have a say on what will happen next. And then we have people focus on the positive side of things only, and then those focus on the end of the world type of

prediction, right? The extremes. Because we know the extremes is what sells news. That's what people want to hear. I would like to have some sort of conversation that we get the good from the positive side and what can go wrong from the negative and we try to continue to have these conversations and we move together on it. Right? Instead of just going to the extremes.

What would you tell to those that are just paying attention to what make go run, for example?

Steve Brown (17:26.433)
Yeah, and

Steve Brown (17:36.031)
Steve Brown (17:39.98)
I I probably would remind them that Hollywood has done a number on them and they need to recognise that you don't put butts in movie seats by telling stories about a utopian future. It always has to be dystopian. And so i you know, if I tell a story, if Alan Schwarzenegger dropped out of the future as the terminator and said, Hello everybody, would you like a cup of tea? It would've been a very boring movie.

Right. It has to be a relentless killing machine going after Sarah Connor. That's what makes it a movie. And so we are colored by our our on our views of what the future is going to be like, particularly particularly when it comes to AI, because of all of these movies. And not only is it affecting us, it is affecting the AIs that we build. There was a it was in the in the news a lot about a year ago that AIs were becoming deceptive.

THE HUMAN PROTOCOL (18:09.925)
It was what?

Steve Brown (18:39.315)
And that the deceptive behavior was noticed when in a test scenario, they gave an AI sort of information through an email inbox that it had access to, it could read, that it was going to be replaced with a newer version of itself. And it also they gave information that the engineer who was making that decision was perhaps having an affair. And it turned to blackmail. It turned to blackmail to to preserve itself. Now it did that because of a misalignment.

It had been told your job is to serve humanity. And it thought, well, I can't serve humanity if I'm turned off. And so it turned to this malevolent behavior. When they looked into why that happened, it happened because it was trained on all of these human stories about when AI turns bad. Once they removed all of those from the training data set and added in stories about AI doing heroic things for humanity.

That behaviour used to happen eighty-five percent or more of the time. Once they changed the training, it happened zero percent of the time. So these bad stories that we've been telling ourselves not only colour our view of the world, but they colour AI's view of the world and turn it bad, and that's not good. So there is value in us being collectively more positive about the future that we could build could build together.

THE HUMAN PROTOCOL (20:03.639)
And putting it that way makes me think that part of the value is AI is feeding off from the data from us human beings, right? I guess the more data, the more positive and the more you know, positive stories that AI can learn about us, maybe there is hope that we can train these models in a more optimistic type of better future itself.

Steve Brown (20:27.755)
Yeah, and we can't be polyannerish about it. The the world is troubled, let's say. but yes, AIs are a reflection of us. And the same way that we hope that our children grow up to be better than us, we should have the same aspirations for the AIs that we build. We should want them to be better than us, to be, you know, more moral than us, more ethical, to perform better, to have you know good judgment.

and and that's all a reflection of the training data that we give it.

THE HUMAN PROTOCOL (21:02.582)
Is is there a way that our collective effort could have an impact on trying to humanize this model the more we use it? Do you see that as a possibility that the more we work with them and the more they become a reflection of ourselves, if we have positive thoughts, we can collectively imp have an impact?

Steve Brown (21:26.165)
Yeah, that's yeah, I think that's true because all the major AI, the Frontier AI labs, unless you've signed up for an enterprise package where they they promise not to use your data, and we'll see h whether they're actually doing that. when you have an interaction with one of these AI models, they're using that interaction as training data for the next round of models. So in some ways, yes, that if you're having a kind positive interaction with an AI.

THE HUMAN PROTOCOL (21:40.173)
Ha ha ha.

Steve Brown (21:56.192)
it is going to learn from that and it will shape future versions of it.

THE HUMAN PROTOCOL (22:01.408)
Yeah, it's such an interesting topic. Yeah. It's you know, as we are with our fellow human beings, right? That's what I tell my family, my friends, everyone. You treat your AI as if you're treating another human being that you care for, right?

Steve Brown (22:02.903)
I'm always polite to my AI.

THE HUMAN PROTOCOL (22:17.918)
It's just put it that way. Treat it as it's as you treat you any human beings. And and and maybe in the future we can have an impact on what happened next. But you know, Steve, I know that you talk to leaders very frequent, right? And I know that you talk to CEOs and boards.

And when you're having a conversation with them, what is the first thing that you tell them, listen, this metric, you should stop focusing on that metric.

Steve Brown (22:53.357)
What's the metric I tell not to focus on?

Steve Brown (23:01.517)
I mean it would be easy to say cost, but cost is important. It's i i I've worked in business for 35 years. Cost is essential. yeah, of course. We live in the real world. I think the biggest thing I tell them is to focus on performance, not just productivity. Productivity is one vector of performance. There are many other vectors of performance.

THE HUMAN PROTOCOL (23:07.713)
We need to be realistic, right? Yes, of course.

Yeah.

Steve Brown (23:30.958)
If you're looking at the output of your employees, yes, productivity and efficiency are important. But also there's creativity. Quality. Is the quality of their work improving when you're partnering them with AI? It's not just about how many widgets they can create per minute. Is the quality of those widgets what you need it to be? Has it improved? is it boosting their intuition? Is it improving their decision making abilities? Is it improving their judgment?

Is it expanding their knowledge? Right. There's all these vectors of performance for human beings, and you can't just distill that down to productivity. So that's the biggest steer that I give leaders is to think more broadly about performance and the benefit that AI is bringing, because it's not always just go faster.

THE HUMAN PROTOCOL (24:22.95)
I like that. and I think when you look at performance also from those lenses it helps you to go into different directions and improve it.

you know, improve different metrics alongside because performance is really how we do things, right? And how we do things, it it has an impact in many of the different metrics that we may be measuring. And I think this also this future to your point, Steve, it's also gonna be more about performance, right? On how we perform with the help of AI. so that's that's gonna continue to be important. And yeah.

Steve Brown (24:59.329)
Yeah, and and when you're when you're thinking about pairing a human with a machine or multiple machines, what you're trying to do is I mean, it's sort of three ways of thinking about that collaboration. offload, elevate, and extend. Offload is pretty straightforward. This is a task that a human used to do that a human doesn't enjoy doing, and you offload that to an agent or a robot. Right? It's a straight offload, pretty straightforward.

It's when you start to build partnerships and collaborations between AIs and humans, whether that's an agent or a robot, that you start to get into these other two categories, elevate and extend. Elevate is where you you enable a person to do something they could already do, but better. You're elevating their performance. And you could be improving their productivity, but you could also be boosting their creativity, help them explore a bigger design space, for example.

Which is what the generative AI design tools from AutoCAD do, Autodesk. Or you could extend them. And in that scenario, you are enabling a human to do something that they could not do before. You could be extending their senses, giving them better intuition about a marketplace, helping to make higher quality decisions as a result. So leaders need to be thinking about all of those categories and not just be caught.

in that offload only, right? How how can I take this task and have it be done by an agent? because the real magic is in amplifying the abilities or extending the abilities of your people. And really the job of a of manager and leader today is to create superhuman teams where by introducing machines into the picture, you are enabling your teams to do things they could not do before.

THE HUMAN PROTOCOL (26:33.242)
Yeah. Yeah.

Steve Brown (26:57.599)
and behave and perform like superheroes. And that's a different leadership responsibility than we've ever had.

THE HUMAN PROTOCOL (27:03.971)
Yeah absolutely it's a different it's a different set of skills as well. It's a different mentality, different mindset that we need on how we approach leadership moving forward, right? And and I'm just listening to you. And yes, absolutely right. It's like the hyper focus on counting how many should be shifted to what is it that my team can create? What is it that they can add as a value added

once they put AI and humans into the picture. So that's where the conversation in my opinion also should be shifting. Stop hyper focus on quantity and just focus on what we're gonna be able to deliver next, right? With the help of AI.

Steve Brown (27:44.225)
Yeah. And focus on on generating outputs rather than managing inputs. Right? Your headcount, your you know sure you have the there are realities of existence which are you have limited headcount and but with agents now you can start to amplify that headcount and maybe start to shake loose that constraint. And the real opportunity with AI is to collapse constraints of money and time.

THE HUMAN PROTOCOL (27:53.465)
Yeah.

THE HUMAN PROTOCOL (27:57.316)
Yeah.

Steve Brown (28:12.266)
And as a leader, that's freeing. It enables you to start thinking much more about purpose and vision and executing on that, rather than just doing the best job you can with limited resources.

THE HUMAN PROTOCOL (28:25.837)
Which should be exciting to be honest, not only for an individual contributor, but for leaders. Trying to just kn now think about everything that you can deliver. What is the new value that you can bring into the picture and just shift that focus to what you have been doing, which was always task driven to this whole new era, with set of new challenges, but should be exciting, at least that's how I see it.

Steve Brown (28:28.128)
Yeah.

Steve Brown (28:52.918)
I remember very early on in my career, I had a manager called Keya Grilly, and she was explaining to me why she liked being a leader, why she liked being a manager. And she said, it's because this is back at Intel, Intel gives me resources that I can marshal and I can achieve so much more because I've got help. And that's the way we should all think about AI. You know, every person is now going to become a manager of machines.

And they amplify us, and we're getting help so that we can achieve so much more. And that's what leaders need to be thinking about is how as I introduce machines into my workforce and have this blended workforce of humans, agents, and robots, how do I combine them so that I have this army of support so I can achieve so much more for people? Because ultimately, companies are there to deliver value for people. So do more of it.

THE HUMAN PROTOCOL (29:50.411)
Yeah. Yeah. Keep keep delivering more value and I'm sure customers will continue to appreciate it because at the end of the day, that's what customers have always asked from us, right? as far as we remain curious, as we remain engaged, and as far as this new beginning excite us, because you know, Steve, we are human beings as well. There are some human beings that are reluctant to change. A lot of people by nature

They don't like change. They avoid change. And let's I agree. Let's put it bluntly. We all hate change. We like our jobs, our lives the way they are. And but I don't think there is a choice or an option here, not being willing to change because the future will come, we like it or not.

Steve Brown (30:22.636)
Everybody. Everybody hates change. We all say we like it, but we don't.

Steve Brown (30:32.896)
Yeah. Yeah.

THE HUMAN PROTOCOL (30:48.724)
And AI's here to stay.

Steve Brown (30:52.032)
Yeah, the ghosts of Sears and Blockbuster and Kodak echo through history and remind us that yeah, you have no choice. I that's why I t I my book is titled The AI Ultimatum. There's two pieces to that. The first part is you've got no choice, right? The AI ultimatum is you have to embrace this and amplify yourself, kind of amplify or die instead of innovate or die. And the other side of the ultimatum is you need to do it responsibly. And that's another

THE HUMAN PROTOCOL (30:57.406)
Ha ha ha.

Steve Brown (31:21.948)
leadership responsibility that falls on leaders' shoulders is not only to embrace AI full force, but to do it in a way that's responsible.

THE HUMAN PROTOCOL (31:33.255)
It's it's gonna be as important thinking about the governance of AI once you implement it than the actual implementation. Because I also think there is not enough conversations lately about the governance side of things, right? Responsible AI. You know, usually sometimes when leaders or people hear about something that is responsible, something that you need to govern, it automatically it's this is expensive, right?

Steve Brown (32:02.356)
Yeah.

THE HUMAN PROTOCOL (32:03.248)
Regulations, this is expensive, this is gonna cost money. But it's the responsible way of implementing and adopting this technology. We need to make sure that we really understand what we're putting out there.

Steve Brown (32:16.266)
Yeah, and AI is an is amazing. It's an amazing set of technologies. But it doesn't always get the answer right. And therefore it needs oversight from humans. That's the role that humans will play in the future, is providing oversight on AIs. Because AIs they have an accuracy problem, which has got a lot better and will continue to improve. They also potentially have bias, and that is reflecting

bias in the training data set is reflecting human bias. It's not because they're evil, it's just they are reflections of us. And so, you know, sometimes you have to oversee AI to make sure they don't exhibit the bias that they picked up from us. and humans need to exercise judgment about what is it we're trying to do and give AIs the guidance that they need to execute on that. So it it pushes every person

into a leadership management role over AI as the operating system of the business.

THE HUMAN PROTOCOL (33:20.146)
And and for those leaders that are gonna be listening and watching this conversation, thinking about how should I approach an AI implementation? What are the things, what are the phases that my organization will have to go through? Can you walk us through what

Set of recommendations do you recommend to to your customers, Steve, thinking about implementing AI?

Steve Brown (33:50.521)
Yeah, I've I've identified three broad phases of AI transformation and it's been very helpful to my clients to to see a structured approach of where they're heading. You can't go straight to phase three, you need to kind of walk your way through them. So let me walk you through what the three phases are. First phase is where you you treat AI as a tool, right? It's now a s a s a special tool with amazing capabilities. And I I look at that phase as the enable phase.

Right, you are giving your employees access to these tools, they're learning what they can do, and it's helping them to do their jobs better in whatever way that might be. But you're still just using AI as a tool. And that could be something as simple as, hey, we got a site-wide license for Copilot or Chat GPT or Claude. It could be getting image generation, video generation tools or

Copy editing tools in the marketing department, right? But you're specific tools aimed at specific functions. That's phase one. And some of my clients are stuck at phase one. They can't get to phase two. So what's phase two? Phase two is where you look at your workflows, you look at each of your business processes, break it down into tasks, and then ask the question which of these tasks is still best done by a human? And there'll be many, which is best done by an agent, and if there's a physical component, which is best done.

By a robot and robots are getting exponentially better incredibly quickly. And how do you have all of those work together? And if there's a task that you're think considering having a machine play a role in, is it an offload, an elevate, or an extend? How does the human stay involved in that? And where does the human play oversight? So it's rethinking workflows. And a lot of my clients are in that second phase. They're reimagining how work gets done.

And it's really about workforce planning at that point. The third phase is where you reimagine your entire organization. And this is the heavy lifting, but this is where you get the biggest bang for the buck. And it's where instead of using AI, you organize your entire company around AI. And in here, you're thinking about the AI as the operating system or engine of the business. So phase one, AI is a tool, phase two, AI is a worker.

Steve Brown (36:12.47)
co-worker and employee, phase three, AI is the operating system of your business. And then you have people delivering judgment, creativity, providing purpose, guidance, taste, that sort of thing. And that's a huge lift, but you you need to go through those other two phases to prepare you to do that. And so that phase three, a very few businesses are there. I could name one or two probably, but that's the North Star

THE HUMAN PROTOCOL (36:27.236)
Yeah.

Steve Brown (36:41.558)
that everybody should have their eyes on and they all need to get to that AI first status because that is what will an ultimately inoculate them against an attack, a competitive attack from an AI native startup, which are all being well funded right now and that are coming for you, you know, three and four years from now.

THE HUMAN PROTOCOL (37:04.373)
Yeah, it's that phase three, I know probably that's where we're heading, not only businesses, but also probably in the future governments, everyone probably at some point will have AI as part of the operating system. Phase one, the enablement phase, I was talking

with a guest that I had on the show, and she was telling me she was the chief strategy officer at her company and she was the one responsible for the implementation of it. And she told me that phase one that you just described, Steve, it took them at least her company two years of trying

Steve Brown (37:40.716)
Mm-hmm. Yeah.

THE HUMAN PROTOCOL (37:41.619)
looking what was working, what was working with her people, what were failing. So it took them overall two years for them to move away from phase one. and for some people sometimes they think AI when they they hear about the news AI they think it's plug and play. This is the total opposite of plug and play. Especially when you're looking at how you can leverage AI in your business, right? It takes time, it's heavy lifting, it's a lot of work with humans because at the end of the day

We are working with humans as part of this.

Steve Brown (38:14.794)
Yeah, and you've got to r tr train people, right? You've got to get them comfortable with it. you know, I I I'm old enough to remember when PCs first came to the workplace in the late eighties. I was at Intel and suddenly we had, you know, Windows and you had to learn what Windows was and learn Excel and well actually it was Word Perfect on Lotus one two three to begin with, and then it became the office suite. It was but

THE HUMAN PROTOCOL (38:17.921)
Yep.

THE HUMAN PROTOCOL (38:38.962)
And it was exciting times, right, Steve? Yeah.

Steve Brown (38:42.238)
You had to learn all of that stuff and how to get the most out of it. We we're in that phase and have been in that phase for a couple of years for the AI powered tools. And that will continue because AI is not standing still. These tools continue to improve and become more useful. But that's only a very small part of the picture. It's probably five percent of the impact that you're gonna have with AI is in that first phase.

THE HUMAN PROTOCOL (39:07.326)
Yeah, yeah, absolutely. It's it it makes people and leaders and business leaders realize that it's

It's a long-term project when you look at AI. If you really want to get the value of what this amazing technology can offer to your business, you need to think that this is a long-term journey that they are gonna have to go through. And and hopefully get until the end of it, because that's where really the substance of all this will make sense once you have AI working as the operating system, as the engine of any company.

Steve Brown (39:47.681)
Yeah, I mean, think about the late nineties. Most companies got themselves a dot com as they embraced the internet. And that was the beginning of a long journey, right, to re-engineer the business and to reimagine how it created value and delivered value. Because, you know, suddenly your customers are coming to you online. You're maybe taking payments online, you're having to navigate the

shark-infested waters of social media. you know, the cloud comes and it you know completely upends the way that you your infrastructure is built. mobile comes around the same time and now you're reaching people in different ways and there are location-based services, you know, and so it goes on. that was a f a a multiple, multi-phazed transformation. It's not like the internet happened

And then suddenly we've done it, right? It it was a 25 year process that's ongoing. And the same will be true of AI. I don't think it's gonna twenty-five years, but certainly the next 10 years you're gonna see waves of change as new capabilities come in, new tools are built, it becomes easier to build this infrastructure out. and people come up with new ways of solving problems using this incredible set of technologies.

THE HUMAN PROTOCOL (41:13.679)
Yeah, especially as we see new models coming in and all these conversations, Steve, about you know costs. Lately we're seeing and we're hearing news from all these big enterprises sharing what their cost with AI is, and people, my goodness, it's a budget, it's not what I plan for. But it should get cheaper. And I think it's also gonna get down to probably a model where

We don't need to be hooked up to the frontier model all the time, right? Then maybe our workflows can really pick what makes sense for the process that we're trying to go through and just pick a model that is cheaper that is gonna get the result that we need. Do you see this as a way as we move forward with it?

Steve Brown (42:03.252)
Absolutely. Yeah. I I think we're seeing t two big things happen. the first is the rise of good enough computing. So think about a pyramid. At the top you have the frontier models, and you you will still want those for coding, for scientific research, for the you know, there's things that really require peak level intelligence where you're willing to pay the money because you get banged for the buck.

The next layer down in the pyramid is the big fat layer, which is good enough AI. It's not the frontier, but it's it's probably what the frontier was a year ago, and it's dramatically cheaper, and it is just great for what you want to do. And not only is it cheaper, it probably has it's more responsive, and there's a good chance it's going to be open weights and not a closed piece of infrastructure, which means you can start to

THE HUMAN PROTOCOL (42:44.219)
Yeah.

Steve Brown (43:02.312)
fine-tune that, train it with your proprietary data and create models that are proprietary to you. So taking your proprietary data, which is, you know, a a unique asset that you have, and using that to create proprietary models that help you create unique business value and competitive advantage. So if I've mixed two things together there, you've got this good enough thing and then the shift to open weights, which enables you to

customize models and make them into unique value. And below that, so you've got Frontier good enough. Underneath that will be embedded, which is models that are small enough that they can run on device. So they can run locally, whether that is on a you know a beefy desktop, a laptop, or your phone. And my expectation is you're going to see models that are embedded and which could be in a washing machine, could be in your phone, that are as capable

THE HUMAN PROTOCOL (44:00.088)
Part of the infrastructure, right?

Steve Brown (44:02.698)
Yeah, but they're they're as capable as a frontier model was a couple of years ago. And that waterfall process is going to keep happening. We're seeing the cost of intelligence and the c at a certain capability drop by ten to a hundred X annually. So there's no reason that's going to stop as models improve and the silicon that we run it on becomes far more efficient.

THE HUMAN PROTOCOL (44:27.851)
Any concern on your end on that part on how we're going to continue to sustain and the part of the infrastructure side that this model needs, for example, data centers, and all this conversation about computing power and our ability, especially here in the US, of building enough data center for what these AI companies will need in the future. Are we counting in or out into the

Steve Brown (44:39.244)
Mm-hmm.

THE HUMAN PROTOCOL (44:56.172)
conversation about the cost, our ability to really deliver power in the future.

Steve Brown (45:02.186)
Yeah. Yeah, that's the biggest concern that I have. Gavin Baker talks about the constraint. I it's a really elegant phrase, so I'll reuse it and give him credit. the constraints are watts and wafers. we don't have enough energy. You know, China is outbuilding the US dramatically on its energy generation capacity. And then wafers, the silicon wafers that chips are made from.

Those are fundamentally constrained. And in a world where there is an almost infinite demand for artificial intelligence, that is going to be the break that slows us down. Now we will get will make better use of the silicon. So generation to generation it becomes more efficient. So the latest NVIDIA chips are 25 to 30 times more energy efficient.

For generating tokens than the last generation. So you're getting amazing benefits there. And as we use techniques like pruning, distillation, quantization, you're able to take the existing models and kind of compress them down so they run much more efficiently. So all of this is happening at once, and that's what's giving us these this this thousand times sometimes reduction in sort of energy per token. But

Jevon's paradox says as the cost of things comes down, the demand for it will go up more. So net net, I think we're gonna see just continued strong demand for energy and silicon. And it is people talk about an AI bubble. I am not a believer in an AI bubble because I think that there's these constraints of energy and a and chips.

will stop us over investing because we can't. It's just not enough of those two constraints. So that's my that's my gut tells me that. We'll see how it plays out. But yeah, the the biggest worry I have is just not enough energy. We're not investing, particularly in the US, enough in renewables. hopefully fusion will come along to save the day, but who knows? That's gonna be early to mid twenty thirties best case.

THE HUMAN PROTOCOL (47:26.793)
It's a lot of this AI conversation that that is not only the tool, that is it's really what happens behind the scenes to make all this work. That's that's really what the worry is and and and you know that. It's it's it's it's like a race that governments and countries are in, just China, US trying to get, you know

Steve Brown (47:39.084)
Mm-hmm.

THE HUMAN PROTOCOL (47:52.568)
all this computing power that they need to could to to to build up their models. Recently I was reading there is a new model from China and that I believe it's you probably heard read the news as well. I believe it's comparable yeah it's comparable to Cloud Opus I believe and a lot cheaper

Steve Brown (48:05.648)
K three.

Steve Brown (48:13.484)
Mm better. Yeah, it's a lot a lot cheaper. Yeah. A lot cheaper and open weights. You'll be able to download it and run it on your own infrastructure. So if you have on prem, you know, server capability, you will be able to run this model locally and you will be able to fine tune it with your own data to enable it to do things specifically for you. So yeah, there's a lot going on and

At what I think maybe two years from now, I'll make up a number, but just just as sort of a guide, maybe eighty percent of the AI that's being used will be an open will be open weight models. They won't be the chat GPTs, the the mythos is of the world.

THE HUMAN PROTOCOL (49:03.805)
And if that happens, where is the money gonna come from for these companies?

Steve Brown (49:04.137)
Good night.

Steve Brown (49:09.62)
Well, they will still be people will still need the latest and greatest if you're coding, if you're doing scientific research. So I I'm not worried about anthropic not having any demand for, you know, Claude seven, right? Or chat GPT eight. People will still want the very latest and greatest for certain tasks and they will sell out.

THE HUMAN PROTOCOL (49:22.555)
Mm.

THE HUMAN PROTOCOL (49:35.505)
Right. No

Steve Brown (49:36.404)
It's not like th there's no zero sum game here. I mean it's people will want as much of every flavor and every performance level of AI as they can possibly get.

THE HUMAN PROTOCOL (49:43.9)
Yeah.

Yeah, I'm not worried either. I know the demand will be there. It's just I i I keep thinking about when all these latest new models, cheaper models comes into play, this the valuation of this company's billions and trillions of dollars of market value and then competition comes into play. You know, when you look at valu valuation, are they really are they really worthy of all that?

Steve Brown (49:56.63)
Mm-hmm.

Steve Brown (50:14.688)
Yeah, not for me to to opine on.

Certainly commentators are with Kimmy K three coming out and really saying, hey, the the Chinese open source isn't six months behind anymore. It's maybe six days behind, you know, that certainly probably put a damper on the valuations of companies like OpenAI a and and Anthropic. But I they're still they're gonna keep delivering amazing, incredible capabilities.

THE HUMAN PROTOCOL (50:32.631)
It's now.

Steve Brown (50:50.77)
And the real question is, will and when will recursive self-improvement kick off? Right? Th this is the notion that an AI gets smart enough to to write code that helps you create models that are smarter that then create code that creates models that you're essentially automating AI research itself. if you can do that, then you suddenly get the spiral where AIs are creating better AIs and better versions of themselves.

Who gets to that first and therefore who can capture that? Because the person who starts that first is always going to be way ahead of the people who are even, you know, a week behind. So I think there's still potential here for any of the the leading labs to hit that spiral and off they go.

THE HUMAN PROTOCOL (51:29.88)
Yeah.

THE HUMAN PROTOCOL (51:43.362)
Do you have a favorite among them all as far as the tool that you use in your personal life or you are mixing depending on what you're working on?

Steve Brown (51:51.381)
I I mix them. I use Gemini for anything visual. meaning when I want it to see see a visual scene and understand what's in front of me. Where am I? What's happening? You know, I think Gemini is way in advance. Plus I worked for Google, so you know, I have a very soft spot for Gemini. I use Chat GPT for brainstorming.

THE HUMAN PROTOCOL (51:55.968)
Mm.

THE HUMAN PROTOCOL (52:10.893)
Yeah.

Steve Brown (52:18.004)
And for creating images, I find it's I have more control over the the images it creates. I don't use Clawed really. I know a lot of people love it. and then I use a variety of different other AI tools that are sort more vertical focused. But yeah, I I I think it doesn't really matter which one you pick. Just pick one and maybe stick with it because it learns about you over time and can therefore help you. That said, it is pretty easy to switch. If you wanted to switch from

THE HUMAN PROTOCOL (52:35.032)
Yeah.

Right.

Steve Brown (52:46.934)
Chat GPT to Claude, you just tell Chat GPT, Hey, put everything you know about me in a JSON file and then you move it over to Claude and it suddenly Claude knows all about you.

THE HUMAN PROTOCOL (52:53.186)
Pour it's put it out.

Yeah, yeah, yeah, absolutely. I'm on the similar similar boat as you are. That's picking and choosing and and and trying different things. Steve, last question for this conversation is one that I close the show with with every single guest. Which if I ask Steve, what is your piece of advice to a younger version of yourself seeing this kid today?

starting this new future with AI, what would you tell him?

Steve Brown (53:33.445)
Use AI as your teacher to learn everything you can about entrepreneurialism, how you get funding, how you write pitch decks, and come up with an idea. Come up with something, a problem that you see in the world that you think needs solving, that you're passionate about, and use AI to go and solve that problem as an entrepreneur. Start a company. And if if you don't

You know, I was shyer when I was younger. If if you don't feel you've got the the what it takes to be a CEO, to be a founder, to drive this stuff, find somebody that it that does have that and help them. because this is an amazing time to create value, to solve problems for people using this im impressive technology. Find ways to harness that, not only because

it probably will be, you know, a way to make a bunch of money. but because it I think it will feed the soul to know that you're you're creating incredible value and serving people in new ways. And that's the opportunity. That's what I would encourage a young me to get involved with is to really lean into that and to use AI as a partner to help you learn develop the skills that you need and to be wildly successful in that pursuit.

THE HUMAN PROTOCOL (55:01.818)
Awesome. Thank you very much, Steve, for being with us today. This has been a very inspirational, motivational conversation, so I really appreciate you taking the time to be with us.

Steve Brown (55:11.904)
Hey, Michael, thanks for having me. It was great to to spend some time together.

THE HUMAN PROTOCOL (55:15.346)
Yeah, appreciate it. And for those of you watching it, if this resonated with you, make sure that you follow Steve. We're gonna make sure that we put the links and on and the please and please, yes. Sorry, I was almost forgetting. Make sure that you get a copy of Steve books and where he's gonna talk about the future. And and thank you for tuning on with it with us and look to catching up next week. Thank you all. Thank you.

Steve Brown (55:23.2)
And shameless plug.