Understanding the linear and exponential deception | Pascal Finette

So Albert Alan Bartlett, who's a futurist, once said that the greatest shortcoming of the human race is our inability to understand the exponential function. And here's what he means by that. We're calling this the linear exponential deception. So the idea of we as humans grew up in a world which is largely linear. So the nature world around— the natural world around us moves on a 24-hour cycle and a 360-year—

Year.

day cycle in the year. It's linear. It's very predictable. 350,000 years of evolution have conditioned us to think linearly. It's really hard for us to intuitively make sense of exponential trends. Now, technology moves on an exponential trend, so it's really hard for us to understand those trends and project those. So let me give you an example. So you take 30 linear steps, so 30 steps one after the other. You have a really good sense how far that is. First of all, you know it's 30 meters or 30 yards. You also have an intuitive sense it's probably like to the door and a little bit further. And you have an intuitive sense how far half of that is. Now ask yourself the question, how far do you get if you do 30 exponential steps, so every step is twice as far? That's close. Very good. So most people don't— so how far is it? I know. See, it's hard. You can do the math in your head. I know, but it's intuitively it's really hard. So you can check in with yourself. So I typically get three answers. The first answer is someone blurts out a mile, which is always wrong. Then you've got the people in the first row who are like they're paying attention and they're like wow Pascal is hyping this up so it must be bigger. And then I typically hear to the moon or to the sun. And then I always ask the The mean question, which is like, how far is it? And nobody actually knows. So to clean this up, 30 exponential steps is a billion meters. It's 25 times, 25.5 times around Earth. It's to the moon, back, and halfway to the moon again. So you were very close. It demonstrates 2 things. The first is, if you're on a linear trend, you move 30 units. If you're on an exponential trend, you move a billion units. In the same amount of time. The second is it's really hard to intuitively understand these trends, which means you continuously make false predictions about the future if you're not forcing yourself to understand that technology moves on an exponential curve. Now, if you plot these things out, there's something interesting happening. So we talked about technology being on this curve. Your thinking is linearly. Now, 3 interesting pieces happen. The first is this here. You think technology is here, but these exponential trends start super slow, and you're disappointed because they are slower than you expect them to be. Your classic example is Google Glass. Anyone had experience with Google Glass? Okay, a couple of you. So I was at Google when we released Google Glass. I was wearing Google Glass for, like, pretty much all my time at Google. And let me tell you, it's too expensive, the battery life is terrible, the functionality is pretty meh, and you look like a complete idiot wearing it. So what you're doing is you're disappointed because you want it to be better. You want it to be this magical experience. The problem is when you're disappointed, you're dismissing it. The amounts of people I showed Google Glass to who said, this is crap and it's not going to go anywhere. It's staggering. But then technology gets better and better, and then you come to the iPhone moment. I was at Moscone Center when Steve Jobs introduced the iPhone. It's this magic moment where, I swear to God, someone next to me, sat next to me, pulled out his Nokia phone, which he just bought, $700 Nokia phone. He looked at the phone and he grumbled to himself, and it's like, OK, I need to throw this away. It's gone. Because we are now in a new era. A phone is not a phone anymore. A phone is a mini-computer. And then you come into chaos and amazement. This is the world where you cannot even keep up with the change which is happening. Kenya's smartphone penetration today is 7%. In 3 years, that number is projected to be north of 90%. In 3 years, a whole nation becomes internet-enabled. So you talk to Barclays Bank, which is the biggest bank, or one of the biggest banks in Africa, they know this, they see the data. And Barclays will tell you, if we are not a mobile-first bank in 3 years, we're dead, because some kid will build a mobile-first bank. I spoke to Barclays about this. A month later, Barclays decided to pull out of Africa. This is the world we're moving into. There's a different way to look at this, which is if you have the acceptance threshold— this is the number where, like, technology gets good enough— People typically do this. Here they say it's worthless nonsense. How many people have you met who've seen like a really bleeding-edge, crazy, but really not sufficient technology? You're saying it's worthless. Then they come and say this is true but quite unimportant. So now they accept it, but it doesn't really matter because it's not big enough. Then they come to this point, and this is really interesting. This is interesting, but a perverse point of view. How many people will tell you that virtual reality is this terrible thing because like you're losing the human connection and you're like in this world and you have these glasses on like you don't see anyone anymore. It's perverse, right? And then you come to the world of I always said so. These are always the people like you meet them in Silicon Valley all the time like, oh yeah, of course, like of course I knew that. So the challenge is if you're staying on this line, if you're staying in your thinking on this line, it's your certain path to doom. This is why Nokia's And Nokia, within five years, went from the largest cell phone manufacturer in the world to not producing a single cell phone at all and making seven hundred million dollars of losses in the last quarter. So here's how this plays out in the world. When I grew up, this was big news. So Gary Kasparov, for the first time, was beaten by a computer in the game of chess. Now chess is easy for a computer to play because chess is a game of like you have about 25 options every move and you basically just calculate a move and then you calculate the next 25 options and the next 25 options for each of those 25 options. It's a brute force attack effectively. It's computationally expensive and it's hard to do, but it's not really hard for a computer to play chess. Now Kasparov Interestingly, barely lost against this computer, which is called Deep Blue. So Kasparov won the first game, he lost the second, they had a draw in the third, and he threw the fourth game. So it's not exactly like computer demolished the humans. If you look at Deep Blue, this was released in 1989, $100 million. Now you know the prices, the numbers. So this thing had 11.38 gigaflops, which by the way is 1/8 of the compute power of an iPhone 5S. The iPhone 5S is so old that you can only buy it on eBay anymore. And if you play chess today, there's not a single human on this planet who can beat your iPhone in chess. The game of chess, forever, is dominated by computers. This gives you an idea of where we're going to. So you take something like Jeopardy! You're familiar with Jeopardy!? It's kind of a weird game. It's basically a Google search in reverse. So, I give you the question, you give me the answer. It's really hard to play. It's really hard to play for a human. Every time I, like, watch Jeopardy, I'm pretty much stumped, and, like, I don't even get the easy questions typically, because it's really hard. Like, you have to have a conceptual understanding. So, this is IBM Watson playing Jeopardy. Thank you for being here. What do you say we play Jeopardy? Let's get right into the Jeopardy round. $400, same category. This mystery author and her archaeologist hubby dug in hopes of finding the lost Syrian city of Urkesh.

Watson?

Who is Agatha Christie?

Correct.

Same category, $600. So this is IBM Watson. It's interesting. IBM Watson is interesting because he plays something which is really hard because you now need to create a conceptual understanding of the world. The interesting piece about Watson is twofold. The first is Watson— so these 2 guys, by the way, are the world champions in Jeopardy, so they're pretty good players. Watson not just like won a little bit, it demolished them. It utterly destroyed the humans. That's the first interesting thing. The second interesting thing is, so Watson happened in 2008, I believe. When this happened, Watson was a room full of computers. You can actually go to the Computer History Museum and see the original Watson computer. 3 years later, Watson was the size of 3 pizza boxes. Today, Watson is an app you put on your iPad. This is where— this is the brink of artificial intelligence. And then you came to AlphaGo. Have anyone seen AlphaGo?

Google?

All right. So AlphaGo is fascinating. I want to play a quick clip about this.

In chess, the number of possible moves is about 20 for the average position. In Go, it's about 200. Another way of viewing the complexity of Go is that the number of possible configurations of the board is more than the number of atoms in the universe. If you ask a great Go player why they played a particular move, sometimes they'll just tell you it felt right. So you can— the one way you can think of it is that Go is a much more intuitive game, whereas chess is a much more logic-based game. We always used to talk about, well, if we could eventually crack Go and have a program that could beat the world champion, then we must have invented some generic general-purpose algorithm. So maybe we're on the cusp of all of that. And we're excited about it, but it is just one rung on the ladder towards solving artificial intelligence.

So AlphaGo is interesting because AlphaGo pushed us into the realm of actual artificial intelligence. And I— if you're building a business today, like, I would, I would pay very close attention to what's happening in AI, in artificial intelligence. A lot of the underlying logic called TensorFlow, Google-based, made open source, so you can actually use it as an API. It's really, really fascinating. So you take this and you put it into movable parts and you get to robotics. So you might have seen Boston Dynamics. This is a company which was acquired by Google and they built these little ones. And when you see the dexterity, the way these things move, And you combine that with the idea of we're actually moving super rapidly into a world where artificial intelligence becomes real, we will come into a really interesting world where we will have these guys around us do fascinating stuff. Who feels a little sorry for the robot? It's interesting, right? So you know how they say anything and everything which is is posted to the internet will be there forever and cannot be unseen. I find this actually fascinating, like when you watch this clip, if I were the guy with the beard, I would actually wear a mask because like one day, like 20 years from now, when the robots will watch that video, right? I mean, they will come and hunt this guy down. I mean, I would definitely wear like a black mask, but— Yeah, he's a mean person. He's a very mean person. But it's really fascinating what's happening in robotics. Now, this is pretty experimental, obviously. This is kind of at the bleeding edge of what is possible with robotics. Let me show you something much more practical. This is a company called Rethink Robotics. They built a thing called Baxter. Baxter is— you can buy this. It's in the market for a while now. What Baxter does is it does 2 things interestingly well. The first is, if you think about robotics and you think about like robots supporting humans in a car factory, when you go to Tesla, for example, there's no humans around the robots because the robots have no concept about what's happening around them. So they would kill you if you would like walk into one of their arms. Baxter has a very clear concept of what's happening around it. So if you walk into Baxter, like he puts on a little frowny face and puts his hands up and basically says like, what's happening? The second is the way you program this guy is by showing him. So instead of writing code, you literally put him into programming mode. His eyes change, so he's in programming mode. You take his arms, and then you show him what to do. Now, Baxter is relevant for a simple reason. Baxter costs you $25,000. So if you're in manufacturing, you can ask yourself 3 times, would I hire someone Like a person who's got, you know, wants to go on vacation and is ill and can only work like 8 to 10 hours a day, or can I buy, would I buy a $25,000 robot which works 24/7, 365, never complains and just does the job? Well, it's an interesting question, right? Like if you think about the future of jobs, like how many jobs are there which are attached to manual labor? So we're moving very rapidly into a world where We will see many, many, many more robots around us.