Google DeepMind Insider: You wont Survive the AI-First Economy, unless... - Steve Brown

0:00Welcome to The Proven Podcast, where it doesn't matter what you think, only what you can prove. Our guest today is Steve Brown, an AI futurist and technology strategist who helps leaders understand how artificial intelligence is reshaping business, work, and competitive advantage. Steve has proven that companies that learn to adopt AI strategically today will define the markets of tomorrow. The show starts now. Hi, Ria. Welcome back to the show. Steve, I'm excited to have you on the show, man. I'm glad to be here. We had a little pre-interview a couple of weeks ago, and I thought, oh, this is going to be a fun one.

0:33it's gonna be fun so for the four or five people on the planet who don't know who you are let's give them a little bit who are you what are the best things you've done uh let me see i spent 30 years or so in high tech good head of hair when i first started not so much now um i spent a long time working at intel when they were the company to be at because the leaders of the world the technological breakthrough people of the world stood on our shoulders um not so much these days sadly. And then more recently, I worked for Google DeepMind. At both companies, I got to be a futurist. So helping them to think about the world 5, 10, 15 years from now and how they want to try and shape it.

1:11And I've been on the speaking circuit now for, I don't know, 10 years or so. I've done over 500 keynotes on five continents, and I help people understand AI and the consequences and the opportunities it presents for transforming business, education, and society more broadly. So there's a lot you talk about there in that one second. We talk about a futurist, and one of the things you've said multiple times is you don't go hide in a room and smoke something and then come out of it. You're actually using... I wish that was what I had to do, yeah. It's so much easier. What does a futurist do if you're not going out there and just rubbing a crystal ball, or at least here in the United States, we had the magic eight ball that you spin around and that thing pops up and says, it may be possible.

1:53What is the difference between a futurist and someone who's just pulling it out of their tuchus? A futurist doesn't make predictions is the first thing. People think that futurists make predictions. What a futurist does is look at trends, how those trends play out over time, and then looks at how those trends will collide in the future to make things possible. So it's about understanding what will be possible in a certain time frame. It doesn't mean it's going to happen. It just means the technology trends are going to the point where this way possible, the people trends, the business trends, and you're looking at the confluence of those at some point in the future.

2:29And you're fundamentally trying to ask and answer two questions. What's the future we want to build? And what is the future we want to avoid? And they're flip sides of the same coin, but they're both important questions to address, of course. Perhaps if we had addressed that when we were thinking more about social media, People thought about the future they wanted to build. Oh, people will share information and recipes and didn't think so much about the future we want to avoid, which is the situation we're in today. Regrettably, people, when it comes to AI, do have a very specific future they're terrified of, that they're trying to avoid.

3:00Most people, when they talk to me about AI, they're thinking, oh, it's ChatGPT and I'm going to be fired and I'll have no jobs and I'm going to be living in a two-story Dorito bag. Let's get the elephant out of the room first and have that conversation. Is AI going to be the extinction of employees? Not in the short to medium term, long term perhaps, and maybe that's a good thing. And we can talk all about that. The opportunity, and it also, my answer depends on how leaders behave. um if leaders stay stuck in a 20th century mindset of well i use technology to cut costs improve productivity and automate as much as possible and that's what i need to do for the shareholders to create long-term value right which is their fiduciary responsibility they need to shift their mindset to instead be thinking how do i use ai to build an ai first company accelerate and elevate the people that I have and amplify the human capital so that I can I stop thinking about growing my company 10 or 20 or 30 percent every year but now I'm trying to grow 10x or 20x per year because I'm using AI as the engine and wrapping my people around that engine leaving scarcity behind because most leaders and managers, they're managing resources.

4:29That is their job, right? They're managing headcount, they're managing dollars, they're perhaps managing some capital and equipment, and they're trying to make the best results they can out of that. If you put AI at the heart of your business, make it the engine, now you can operate in this world of abundance. It's a different way of thinking that most leaders haven't got yet. If they stay stuck in that 20th century thinking and they just automate away the people, yeah, it's not a happy ending. But if we use AI to amplify the impact that organizations are able to have and to amplify people and to do more rather than thinking about how do I do the same with less, then medium to short term, I think things look pretty good for employment.

5:19Well, I think one of the things you just brought up there is important because you come into organizations not to teach people how to use chat GPT, but you teach people how to fundamentally change how they run organizations with an AI first model, because it's not a debate anymore. It's not like this may or may not happen. You know, one of the examples, and I still quite a few examples from you, and I will give you credit where credit is due, is you talk about that the robots that are made today and AI of today is the worst they will ever be. They're only going to get better going forward. And one of the examples you gave that I just fell in love with was there was a, I'll let you do it, the medical one that went from 10 days to two.

5:56Can you tell a little bit about that one? 10 years to two days. Yeah. So this was an AI from Google called Co-Scientist. And it's to help with scientific conjectures and trying to determine why something you're observing may be happening, what's the explanation for it. And one of the things that stumped scientists for quite some time was figuring out, in a hospital environment, where you get things like Mercer and other things, where you have these rigorous cleaning protocols, but you can't get this bug out, right? And people are still getting sick from it. why might that be happening? What's the phenomena?

6:39And it took scientists a decade to come up with this idea that maybe things like Mercer, they're forming chains, they're linking into chains, so they can actually crawl off surfaces and onto other surfaces. It's a bit like prisoners tying sheets, bed sheets together to get out of their accommodations, let's call them. same thing they were they were just using the forming in chains and then moving around and when they they presented this information this challenge to google co-scientist it thought for two days and came up with the same answer so you know you go from 10 years of humans thinking about something to ai being able to come up with it in a couple of days and you see many examples like this particularly in scientific research and discovery where another example from google i was at DeepMind when we were working on AlphaFold and doing the release of that to the scientific community.

7:38And that crashed the understanding of the structure of a protein. Normally it took four or five years. Basically it was a PhD of one person spending four to five years of their life trying to decode the structure of one protein. There are about 20 ,000 different proteins in your body. And it was now able to do it in seconds, accurately. And that information is now empowering the medical research industry for the next 10 and 20 years. And we will see new drugs and new therapeutics come out of it, perhaps understanding disease paths for Alzheimer's, Parkinson's, and other debilitating conditions.

8:19And we'll see breakthroughs in the next five to 10 years around that because this AI was able to do all this decoding. So this is just the start of what's about to happen. And the reason I ask is we turn 10 years into two days. You turn four years into seconds. And the reason I bring this up is if you're in business, if you're trying to scale, if you're trying to get moving forward, this is what's already coming. This is the tidal wave. You're already sitting on the beach. You're going to have to survive this. You're going to know how to serve this and do this. And people are looking for tactical ways to do this because we're in different stages of AI for, and you said it really well, we've had AI for a really long time.

8:55This isn't anything new. AI has been around for an immense amount of time, you know, spam filters, that's AI. That's just, it is what it is. When you're sitting there and you're looking at a business owner and you want them to be AI first, which is a completely different mindset, much like, cause I'm old. When the internet came out, there was a whole pivot that we had to do. We're in that same pivot. One of the tactical things that you're sitting down with your clients of the people you work with say, okay, this is where you are. It is what it is. The roads used to be made out of concrete. They're now made out of water.

9:24You're going to have to change how you show up. It just, there is no way around that. What are you sitting down? What are the first kind of four or five steps you say, okay, this is a reality. Let's start doing this. Yeah. Great question. So a lot of my clients, you know, I show up and they, they want to pat themselves on the back a bit and they'll say, you know, we, we, we have an enterprise license for co-pilot now. This is going to make a difference. And I said, well, good, that's good. That's great. But that's not, you're 1 % of the way there. The other 99 % of the opportunity is still on the table.

9:59So let's talk about what that is. And the first thing I do is start off with a thought exercise. Imagine an AI native company that is coming after you. They're well-funded. They have the people, the talent that they need. They have access to AI technology. And once I've explained the different types of AI, because there are many different flavors, as building blocks, like Lego building blocks that they can start to put together to make business solutions. Once they understand those, I say, well, okay, if you're an AI native competitor, if you left this business tomorrow and you went to this AI native company, what would you do to compete aggressively?

10:41What does that AI native business look like? AI native being you're building a business from scratch, knowing everything that you know about AI and knowing about its capabilities now and in the next two years. What would you build? Would it be the same as the business that you've just come from? Well, no. Okay. Well, let's talk about what that would look like. And then how are you going to start to build that inside your existing company? How do you become, You're not going to become AI native because you're not, but you can become AI first. And I give them the example of internet first businesses.

11:18Amazon, great example, right? They were an internet first business. They came after retail and started to eat retail's lunch. So what did retail do? They became not internet native, but internet first. and Target and Walmart and all of the big box companies that Amazon was coming after adapted and said they invented this whole omni-channel idea, right? Where you could buy online, return in store. And their focus was on, yeah, we're going to match Amazon capability for capability. We're going to go online and do all the things, but we're going to leverage the fact that we're not internet native.

12:01We're internet first. and that brings us some capabilities that amazon doesn't have and the mantra in the industry i used to do a lot of work with retail at the time was wacd what amazon can't do they don't have a physical presence and now they've bought whole foods and ironically they're now going to have big box stores of their own and go off to target um but there's there's something about using the physical value you have today. There's something good about the businesses you have. You don't feel bad about it. But how do you then embrace AI to become AI first? And that should inoculate you then against an AI native attack.

12:43And that's the way I start them off. So how does one do that? How does a company become AI first and do that pivot? Because yes, Target and Walmart and all of them adapted and they did have an internet presence. And this This is what we can do that other ones can't do. And other ones didn't, like JCPenney's and Sears and Kmart. They were too late to the game and they don't exist anymore, which is still wild to me. But it's going to happen if people don't adapt fast enough. What I'm trying to do here is give people that competitive advantage. And I think that's what you try to do as well. Absolutely.

13:14And it means you, first of all, have to let go of the past. You have to let go of today, the status quo. A good example of a company that didn't do that was Blockbuster. Netflix used to compete with them with DVDs. And then Netflix realized, oh no, we need to become a pivot to being an internet first company. And we're going to let go of mailing people DVDs, go online. We're going to upset some customers, but we're going to leave these blockbuster people behind. So that's what you're trying to do is think, given the new technology, how would I deliver value in new ways? What are the approaches I need to do that.

13:52And so to become AI first, there's three chunks of effort you need to go through. The first one is, how do I use AI to amplify my people? And you want to ask questions like, if my best employee could be everywhere all at once, what would that unlock? Because that's what you're trying to do. You're trying to use AI to amplify people, to use AI agents to mirror the abilities of your human talent and to amplify them so they can effectively be everywhere at once. So that's the first sort of question you want to ask. If I could use AI to offload all the busy work from my employees, what would that allow them to do that they don't have time to do today?

14:37So think about how do I use AI to unlock the latent abilities and capabilities of my human talent, not to replace them, but to amplify them. And with the goal of how could I have 10x the impact we have today, let's say a year from now or two years from now. So that's the first bucket of effort is amplifying your human talent. The second one is - Can you give us an example? Can you give me an example of that before we lose people? Like someone you've worked with, like this is what we actually did. um it might actually be too early to give a good example of this um just because agent technology is what we're really talking about here and we're in the early phase of agentic ai um and agentic ai is essentially it's a smart piece of software that that behaves like a digital employee.

15:35And they're going to get better and more capable over time. But what you're trying to do is pair your employees with agents. And there are going to be three types of agents, by the way. So let's talk about those and how you might put them into your workforce. The first type of agent is an offload agent. It does exactly what you'd expect. It offloads a task from your employee. This is the freeing them up from the busy work, booking meeting rooms, booking flights, that kind of stuff. So you're offloading. The two more interesting types of agentic AI are elevate agents and extend agents. Let's talk about the difference between those two.

16:15Elevate agents partner with a human being so that you pair them with them and they're collaborating to elevate the performance of the person. Now notice I very carefully chose my language. I didn't say elevate the productivity. That might be one vector, but elevating the performance, meaning their productivity, their efficiency, their effectiveness, their intuition, their creativity, their decision-making abilities, and so on. So you have to think multidimensionally there. So how am I elevating the performance of my employees by partnering them with agents? And then the third category, more rarefied air, but if you can get there, this is amazing, partnering a person with an extend agent that extends the human's capabilities so they can do something they could not do before.

17:08So let me give you a concrete example. This is not agentic technology, but at least gives you this idea of amplification, expanding creativity and productivity. Generative design technology. This has been around from Autodesk for some time. What it does is it takes the initial design that a human designer has created and riffs on it. And it comes up with hundreds or thousands of variants of that design. And the human designer said, hey, I'm trying to optimize for this or that. So a good example I talk about in my keynotes is the bulkhead design of an A320 aircraft. Now we're getting super concrete.

17:49The skeleton that's inside that bulkhead is this weird design. And when you see it, you think, well, that wasn't designed by a human. And it wasn't. But it also wasn't designed by an AI. It is the result of a human and AI collaborating together. What the human does, they create the initial design. They specify to the generative design software, which is AI-based, that I'm trying to, it needs to handle four and a half G's of tension, right? Because the bulkhead is there to hold the fuselage together, kind of important. But it also matters on a plane what the weight is. So there's a reason that we don't have peanuts on planes anymore.

18:32It's because pretzels weigh less and the airlines figured that out. United Airlines did a calculation that if every one of their passengers took Ozempic and lost 10 pounds, it would save them $80 million a year on fuel. So weight matters on a plane. So this generative design software then optimizes the design. It sort of riffs on it, comes back to the human designer and says, I put these all through simulation. The one that you want is this one because it optimizes for weight, cost, which is always a factor, and it still gives you the tensile strength that you need to hold the plane together. Now, what's happening in that scenario?

19:08A human designer might have the luxury of the time to come up with three to five options and then run them through simulation and then okay we'll pick the best one using genitive design now you can explore a much broader design space you can look at a thousand or two thousand or ten thousand options what you're doing is collapsing time and that's what you're trying to do in the second cache that i'll come in a second, but you're using AI to one, get you to a place you could never have got before. The designer would not have come up with this thing on their own. So it's somewhere in this elevate and extend area.

19:51But the AI would not have come up when it's on its own either. It is the collaboration of a human and an AI working together. So companies that use these types of tools to expand the design abilities of their engineers, for example, are able to explore much broader design spaces much more quickly and essentially collapse time and money. And that's the second thing that organizations need to do is how do you, as leaders, how do you create abundance in your organization? How do you create more time to do things by collapsing cost and time, reducing marginal cost of design, for example, but in any other realm so that you can free people up and get to the third stage, which is stop defending your existing business and start thinking about how do I develop new businesses to go serve new needs in the marketplace?

20:54I think right there, stop defending your existing business is important because I think the JC pennies and the Kmart's defended like, no, this is who we are. We're going to double down on this where the Kmart's and the targets and the Walmarts were like, nope, we have to change. Just, it is what it is. The rules have changed. Amazon's changed the rules. We need to, we need to change that for most business owners though. It is still like telling them, okay, I need your, you're a two-year-old. You're a toddler. Congratulations. I need you to dunk a ball. They're like, what the heck? They're begging for something proven.

21:24They're begging for something tangible that they can turn around and they're like, okay, well, I'm going to ask my 12-year-old what chat GPT is. I'm like, that's not going to work. you know so having that conversation telling them is it a is it a course is it a material is it where can they go where can they go into this idea and start learning and playing and understanding because i think we're at the point a bit like climate change where things are changing there's just nowhere the science points at it um you know if you if we want to be somewhat politically correct like no we didn't necessarily cause seasons but we definitely gave it a red bull to speed it the hell up but people are like well we don't know what what to do and i think it'll go into the idea of okay i know what i need to do but i what i'm not willing to do and that is a completely different conversation as far as the not willing people who are listening to this really simply you don't have to don't worry about it you'll be bankrupt you'll be gone you'll be you so don't change but for those who actually want to change they're looking for something tangible and i hear this all the time when i'm on stage or if i'm talking to people organizations owners business funders the people you interact with the people that bring you in and say listen we know this is coming there is a tidal wave coming how do we do this right now how do we educate our senior staff how do we educate our junior staff how do we survive so this is this is the need i identified a couple years ago it's why i created an online course for senior business leaders to help prepare them for this time and i wrote a book the ai ultimatum again designed for leaders to prep them on what are the different moving parts of AI?

22:56What are the different flavors in there? What is agentic AI? How do I deploy it? What are the things I need to worry about? How do I deploy it successfully? What are the gotchas that you can have? Why have 90 % of AI deployments today been judged a failure? Well, it's because XYZ, and I go into that and explain what you need to do to have successful deployments. What's the role of physical AI, robots, in your operations? If you have a physical component to your operations, and most companies do, unless they're a pure play SaaS company, you're going to have some sort of physical element. People are shocked at how fast robots have matured and how quickly they're coming into our lives.

23:43You know, people will be selling robots into homes, humanoid robots into homes this year for$20 ,000. You know, a lot of people haven't really figured that out yet. So it's coming much quicker than they realize. The other thing I try and help people think through is that you have to, you know, you mentioned earlier that one of my mantras is AI is as bad as it's ever going to be. Robots are as incapable as they're ever going to be. They're just going get better and better from here. There is an exponential curve that we're on. And exponentials, when you look backwards, look linear. And then there's this wall ahead of you, and you have to understand that.

24:23The example I give people is back in the day when Gmail was launched. People thought Google were nuts for giving people a gigabyte of free storage with their Gmail. And they're going to put themselves out of business. They can't possibly do that. It's going to bankrupt them? Well, because Google realized that they're on an exponential curve and a gigabyte was going to be nothing, you know, in five years time. So you have to understand this exponential nature and plan accordingly. I think that's what we talked about in the very beginning. What took 10 years took two days. And, you know, there's a lot of people who have resistance to having robots in their house.

24:59And we have this conversation all the time, like, oh, I'm not going to have a robot in the house. I'm like, you see that little hockey puck that vacuums your house? that's a robot so it just it's already happening and we just i don't i think that's one of the things that having listened to your keynotes and having read your book you talk about how ai is not new we've been with it for a while we talk about the junk mail and all that can you give examples to get people to kind of exhale a little bit because they're terrified of it right now what are some of the examples of ai that we've already had for some time well every time you open your email there is an ai that sorts the spam from your inbox and it's just making predictions based on having been trained on literally billions of emails.

25:38It knows the signs of spamminess and it makes a prediction about whether it should be put into the junk folder or your inbox. And it's good. It's accurate probably 99.9 % of the time. So you use that every day. Let me give you some other examples. When you post a letter and you handwrite a zip code on there, the post office for 15 years has used AI, a convolutional AI, that was what it was invented for by Jan LeCun back in the day, to read your scroll and turn it into numbers. So this sort of stuff has been around for a long time. I'll give you one more example. In the retail world, if you're a chain store in apparel, you have to decide what products are showcased in which stores, in which locations around the country or around the world, if you're global, and what colors in what sizes of what product lines am I putting in each store.

26:35And you're trying to optimize for sales out and you're trying to minimize returns and stuff moving into the bargain bucket, right? And being sold off for pennies on the dollar. The reason AI comes into play here is it's optimizing based on historical sales out data. And it's understanding innately that people in different parts of the country have different tastes and they're different size profiles. You can't just have a bell curve, that bell curve shifts depending on where in the country you are. And so it's called assortment planning and it's been used for 15 years. So, you know, AI in the form of machine learning to optimize business processes, it is running in the background of global business and has been for 15 years what's new is the generative ai stuff and that's what's got everybody's attention but ai has been around a very long time it's interesting you bring that up regarding the the size of clothes and all that one of my favorite ones of this is ironically target target would track using ai and using algos what you were buying they're like oh my god this person obviously is pregnant based on data and they would send something say congratulations you're yeah you know this example you're pregnant This is what's going to happen.

27:47People would freak out. They're like, Target knew I was pregnant before I knew I was pregnant. So their solution was they just lied to you. They sent you the same stuff. They just changed the catalog they sent to you that had random other things so that you're like, oh, look, it happened to have this in here because Target knew that. So having the idea that this is based off behavioral science, this is based off behavioral economics, AI is just leveraging all those tools. It's already been happening for a really long time. If you're just waking up to this, you're late to the game. So what are some of the...

28:15Yeah. So what are some of the industries or jobs specifically that you're like, no, it's over. You need to retool right now. And then we'll talk about how to retool in a second. I mean, the most obvious ones that are going to fall, let's say, in the next 10 years. Let's be generous with our timeline. That's very generous. Accounting is probably a big one.

28:45Driving is probably another. whether that is truck driving or you know taxi driving that's likely going to evaporate but they won't go completely there'll still be truck drivers who are driving in and out of the city because people feel more comfortable that way but they'll drop it off at a hub out of town it'll do the freeway journey a robot will do that and there'll be still be some taxis around because some people will not want to go into a self-driving car, but most people will do it because it's going to be one third of the price. So, you know, eventually we'll get, we'll get, we'll get over it and get over our worries and do it.