Strategic AI Integration in Engineering Teams: Lessons from Google's Applied ML Approach

Jod Kaftan at UXDX USA. Video: https://youtu.be/hGqHaV3ePMU

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

Why designers should design futures

[00:00:01] Charles Eames said design is the expression of a purpose. When you think about that, for me I interpreted, everything can almost be designed except nature. So organizations can be designed, policy can be designed, governments can be designed. The Declaration of Independence, you could say, is a design artifact of sorts. But more importantly, futures can be designed. And the question is, why haven't designers gotten involved more in designing futures? Well, we've kind of always done it in some form or fashion. So a little bit of what I'm going to talk about is actually the vocabulary and methods and processes from both futures thinking and foresight. But here's the thing: if designers don't get involved with designing futures, with our empathy and our creativity and our experience, who will?

[00:01:04] So, content. I'm going to talk a little bit about the theory behind foresight, and you'll see we're going to talk a lot about the evolutionary process. And part two will be more practice, so some interventions you could maybe try when you get back to the office.

How our brains relate to the future

[00:01:26] The thing is, we need to understand what futures thinking is and why it's relevant to design. I'm going to spend a little time on this slide because we need to understand how our brains relate to the future. Our brains are neurologically wired to predict, protect and imagine the future. Evolutionary selection has adapted us to a process called active inference. This is a probabilistic method in which the brain is constantly trying to predict, anticipate and imagine the future while minimizing prediction error. There's also the salience network in the brain, which is constantly scanning the horizon for threats, which probably kept us from being eaten by dinosaurs.

[00:02:30] But more importantly, there is a theory here too, it's called dual process theory, and you could think of it as a dual operating system for the brain. Some might call this the emotion, cognition, action cycle. These two systems: the first system is fast, it's intuitive, it's emotional, it's subconscious, and it uses 95% of our brain's activity. The second system is slower and it's rational, logical, uses the prefrontal cortex and problem solving skills, and that uses about 5% of our brain's activity.

[00:03:21] This is the cool part. The second system, its whole purpose is to skillfully choose among the competing concepts in the first system, to be more reliably accurate about predicting and anticipating the future. I know that was a lot of science, but my point here is that futures thinking is part of the human condition, neurologically, biologically, and it's tied to the evolutionary process. We're doing it anyway, so we may as well do it consciously. That's what it's about. Our brain is always trying to predict the future.

Foresight and futures thinking, signals and drivers

[00:04:07] Just to untangle semantics a little bit here, it would be irresponsible if I just conflated these two things. You could think of strategic foresight as applied futures thinking. You can practice futures thinking without doing foresight, but you can't practice foresight without futures thinking. Foresight is more traditionally tied to strategy. It uses both quantitative and qualitative methodologies, and it's really also interested in probabilities and possibilities. Whereas futures thinking, it's a little more art school in turtlenecks. It's evidence-based, referenced largely from signals, which we'll talk about in a minute, but it's more predisposed to possible futures, and really trying to imagine and speculate those futures.

[00:05:16] The interesting thing is, strategy and foresight used to be like best buds. And what happened? There could be a variety of things. One theory: the short-term tenures of CEOs, who don't really care about what happens after they leave and they hit their number, created this divide, and the pressure of short-term revenue pressure created that divide. But the crazy thing is, without foresight, strategy is prone to disruption. Without strategy, foresight is inactionable. So we have detangled that a little bit. I'll just use foresight from here on out, for the sake of simplicity.

[00:06:14] Okay, so now we have signals and drivers. The best definition of a signal is this, and it's my favorite definition: a specific example of the future in the present. A signal is something that you can only get to through drivers. Now, what's a driver? A driver, metaphorically, is like a tide. It's a broader trend. So something like, how will AI and robotics replace jobs in healthcare 10 years from now, would be an example of a driver. But you can only get to a signal through a driver. The signals are sort of like what gets reflected in your scenarios, which we'll talk about in a second, and an example of a signal related to that would be, well, ChatGPT just passed the medical licensing board exam. So signals and drivers. A signal can be a breakthrough discovery, it can be a new policy that comes out, it could be a new product, something that's going to impact the way we live, work and go through life.

[00:07:40] Let's look at a few examples. If the driver is, city planners are looking for a way to understand climate resistant infrastructure, let's say that's a driver, the signal would be that the United Nations approved designs for a floating city that withstands category 5 hurricanes. If the driver is that scientists are trying to figure out how to prevent overfishing of our oceans, the signal would be that cell cultured sushi is going to market, so seafood that's basically created in labs.

[00:08:30] And then finally, if energy industry folks are looking at things like microgrids as a trend in the future, the signal would be that there is a camp for the unhoused in Berkeley, California, that had a large solar panel donated from a private donor, and when the camp moves, all that energy infrastructure moves with it. And therefore you have a signal that raises questions around the mobility of energy infrastructure.

Positive and shadow imagination, and scenarios

[00:09:09] There are two lenses, and there are many other lenses, but I like these two for their simplicity. As you can probably guess, positive imagination is like applied optimism. It's the future we want to live in. And so you can then work from signals through scenarios, which we'll talk about in a sec, to imagine that world you want to live in. But then you would think shadow imagination, which sounds dystopian, and yes, I guess it could be. Practicing shadow imagination allows us to be better, adequately prepared for unforeseen challenges.

[00:10:00] In fact, about 14 years ago the Institute for the Future ran a massive online simulation game, role playing game, and basically some of the assumptions in the game were that there's a massive pandemic, and media was decentralized and no one knew where to get the truth. So people played this game about 14 years ago, and then when COVID happened they surveyed the people that played that game 14 years ago, and they scored much lower in terms of their anxiety when COVID hit, and they felt more adequately prepared with how to respond to that. Which is so cool, because what they did is they pre-processed the challenges in their brain, so their brain already had something to kind of build a relationship with and have an expectation around. So really interesting data there.

[00:11:04] Scenarios are basically the expression of that possible future that foresight practitioners and futures people use. You've already been doing them. I have some friends, we talk about this stuff and we're like, it's very similar to design, and you're going to see that as we go through this. If you've done a user journey from the perspective of your user, it's a flavor of a scenario. But in futures, scenarios could be a video. Like Microsoft made one, also like 14 years ago, I feel like, and it was a video where this woman has this wearable and she can wrap it around her wrist. Does anyone remember that? And then it turns into a phone, and then she's on her mirror. At the time no one had thought of that, now it's commonplace.

[00:12:00] So that can be an example of a scenario. Or a scenario in the function of a workshop activity could just be writing out a 150-word story. But those scenarios need to be credible, and to be credible they need to be informed by signals, by real strategic inputs. That's why in futures thinking we call it evidence-based imagination. And when I saw that word imagination, I was like, I'm going to learn more about that. Okay, so that's scenarios.

Evolutionary development and the double diamond

[00:12:37] So you could see evolutionary development. Human creatures have mastered complexity through the evolutionary development process. Evo Devo describes the positive friction between the evolutionary, which is emotional, it's generative, it's creative, it's indeterminate, and it diverges, and the developmental, which converges and is constraint generating, stabilizing, and so on.

[00:13:25] Now, interestingly enough, this kind of blew my mind, our genes are doing this all the time. 95% of our genes are evolutionary. They're mutating, they're seeking permutations and possibilities. And 5% of our genes are developmental, and what they're doing is functioning as cat herders for the other 95%, and basically they are stabilizing the successful adaptations that were made by the 95%. I was talking to someone about this the other day and I was like, it's almost like the developmental part of it is like a creative director, you know what I mean? They're getting multiple designs and they're like, I want that one, I think that one's going to give me a more resilient future. So that's what your genes and your brains are doing all the time, actually.

[00:14:29] Good, now we have a little build here, I think. Ah, it didn't work. This is what happens when you upload something to Google Drive and then download it, some weird pixie dust happens. But anyway, all this diverging and converging, I was like, wait a second, where have I seen that before? And the double diamond comes to mind. So what does it mean? Maybe the design process itself is very faithful to the evolutionary process, and that's why it's so effective. Because if you can think of one system that works, it's evolution.

Hindsight, insight, foresight and the types of futures

[00:15:11] So let's talk a little bit about, oops, I went too fast, sorry. So, looking at that Evo Devo cone through a temporal lens now, you have hindsight, insight and foresight. When we look at hindsight, which is the past, we can learn about the developmental choices that were made, so that when we do go into foresight and explore the future, we don't make the same mistakes, and we can learn from that, and so we have more successful foresight exploration.

[00:15:56] When we go and do hindsight and foresight and we bring it into the present, that constitutes insight. And the insight is important because in the present is where all action exists. You can't change the future if you don't do it right now. So basically, to be effective at this you have to look back, you have to look forward, and then you have to come back to the present.

[00:16:29] Now, if we flip our yellow cone around, you can see that there are three types of futures. There are preferred futures, I'm sorry, let me start from the bottom. There are probable futures. Probable futures are informed by hindsight. There are possible futures. Possible futures are informed by foresight. And then, interestingly, preferred futures, which only come into being after looking at both possible and probable futures, are informed by insight and action and values and choices that we make about what future we want to live in and what future we want to avoid.

[00:17:19] In fact, you could go all the way back to Aristotle on this. Aristotle talked about the good, the beautiful and the true. The true is like science, it's discovery, it's logical. The beautiful is art, it's creativity, it's imagination. And the good is judgment, it's moral imagination, it's values. Same thing.

[00:17:47] Now, this syntax of the three futures came from Alvin Toffler, who wrote the book Future Shock. Highly recommend it. But the interesting thing I want to just leave you with here is that for a preferred future to be true, he says it has to be political. For those of us that don't like politics, futures is a tough game, I'll tell you. But there's no other path through realizing the future you want outside of politics.

[00:18:24] Okay, so now this is an evolution of the three Ps, and this is what's called the modern foresight pyramid, it's the four-P pyramid. Really the only difference here is that in the upper left they've added preventable futures. Preventable futures are there because all living systems with complex brains are all regulating their behavior between pain avoidance and pleasure seeking. So by adding preventable futures we are able to practice in such a way that we are more faithful to the duality of our emotions.

[00:19:13] And one thing to point out here, predictive contrasting versus sentiment contrasting. Predictive is really trying to determine what events will happen and what events should happen. And then sentiment contrasting is more about what values will prevail in leading us either to a future we want to avoid or a future we want to see happen.

The six domains and the loop

[00:19:43] Now we have the six domains. Really quick on this: like anything, when you're telling a story, if you only think of it through one perspective, you're inevitably going to have an unintended consequence, you may have bias. So when you practice speculating the future, you should see it through these six domains. And again it's like a scaling. Imagine if Facebook did that, and some other missteps, and they thought more about the social consequences of some of these product releases. Everyone learns in retrospect, I hope, but that is the point of the six domains. And again, following them helps you to create more resilient, more credible futures when you're trying to implement them.

[00:20:46] Finally, there's this notion, it's called, yeah, something happened to my deck, it's called the Do Loop, that futurists refer to: learn, foresee, act and reflect. And I can't help but look at this, and I was like, wait a second. And then I thought about Eric Ries's think, make, check. And then of course, yeah, that was supposed to move over but that's okay, the design thinking, or human-centered design process, is very similar. So you can start to see a lot of similarities, and when you get into this stuff you're thinking, why aren't designers designing futures more often?

Practice: barriers to long-term thinking

[00:21:39] Okay, let's get into some practice now. These are two of my favorite design quotes. We've already talked about Herbert Simon and preferred futures, but I really like Christopher Alexander because it's really important. Why? I don't care if we get invited to have a seat at the table. We're at the table. And there's one thing that we can do really, really well, and that's bring form to where there is abstraction. We take the intangible and we make it tangible, we make it emotional, we make it something you should believe in. So absolutely, this plays to our strengths.

[00:22:23] Now, how do you get started? These are just some common barriers to doing anything other than short-term in your organization. The problem with short-term revenue pressure is you have these CEOs, like I said, on short-term tenures, and they don't really care about what happens after they're gone, and we just need to recognize that's a reality. We'll talk about ways to address that.

[00:22:55] And then managing uncertainty and balancing and managing risk and all of that, that's been taught in MBA programs for decades. We can understand executives being risk averse. That's their job, and it keeps the lights on, it keeps everyone paid. But what we can't easily overlook is being feckless and lacking in courage, and you need to have long-term thinking in order to do that.

[00:23:23] And finally, existing processes. This is the hardest nut to crack. When you want to try something different in your organization they're like, well, we've always done it this way. It's a very hard kind of orthodoxy to challenge. So we'll talk about some ways you can do that, but basically if you can find a way to just nudge it a little bit without disrupting those processes, you can have a greater chance of success in changing the culture.

[00:23:52] These are just some very cherry-picked things. This in no way represents the entire toolkit of things you could try, but there are a few things I wanted to just cover really quick. How am I doing on time? Oh, okay, I'm sorry. Well, I think I unfortunately used up the clock. You can see these, I think the decks are available to everyone, so you can download it, there are links in there, all these things you can try. Again, thank you for having me. Thank you.

Q&A

[00:24:41] Host: I love that we can control the future in our brain, I love that concept. Let's see what questions the audience have. Great session on foresight and design fictions. Can you walk us through a real life example where it has been applied in a business organization, and what is the outcome of it?

[00:25:05] Jod: So there are little bits of examples that I can refer to, and I can even tell you about things I've tried where I work. One of the things I was going to talk about was this concept called a futures wheel. People use this all the time. Royal Dutch Shell used it in the 70s to anticipate the changing of their business and how the landscape was changing.

[00:25:38] Jod: But I just did a futures workshop about a year ago with energy executives, and I thought this is going to be risky, they're going to be like, what the hell is he talking about. But I've never seen a more engaged group, because the thing that made this fun for them, and again, fun and creativity by the way have a lot in common, one creates the other, they had, because there are no facts in the future, and because we use signals, they actually came up with some really cool ideas. Like a digital twin where you could have services and people could pay their bill through virtual reality. I mean it was all this weird stuff, but the point was it engaged an industry that is highly volatile to anticipate and think about the future.

[00:26:30] Host: Great. All right, does Oracle use any of the futures foresight methodology to inform strategy or vision in the next phase of the product?

[00:26:41] Jod: Well, I'm working on it. So where I work it's still a little bit siloed, so I'm a little bit of an oddball. A couple of people from my job are here. But you know, it's very siloed. Strategy is kind of like way over here and it's on this ivory tower, and here's the strategy, can you activate it, R&D? And what I'm trying to shift is, let's co-create as we go, together.

[00:27:13] Jod: So one of the things we've done, it's not necessarily futures, but to address that, is we did the sprint where we're parallel pathing and strategy is prototyping with us. A lot of people don't realize strategy can be a prototype. So that's where we want to get to, and when you have that kind of coalescing of collaboration, and people having understanding at the same time, incrementally, you have a much better opportunity to try weird stuff like this.

[00:27:44] Host: Definitely. All right, when you introduced this framework to Oracle, have you encountered any barriers, and how did you overcome it?

[00:27:55] Jod: Surprisingly, this kind of started when I looked at our conference and it was kind of like AI, AI, AI, and I was like, what's missing? So often I try to counter the sort of monolithic predictability of the things they want to talk about. So if we're focused on AI, I'm going to be focused on human-centered AI. So I just pitched, like, hey, what if we did something with our clients and customers with all this uncertainty about energy futures? And they're like, yeah, okay, we needed a speaker anyway. So just suggest something, you have nothing to lose.

[00:28:37] Host: That's awesome. All right, I love that, leave us with, just suggest something, you're never going to lose anything, might as well try it. That's great advice. Thank you, Jod, so much, appreciate it. Really informative, pleasure meeting you.