The a16z Show - a16z Podcast: Not If, But How -- When Technology is Inevitable (with Kevin Kelly)

Episode Date: June 7, 2016

Technology has always been a force in how we live, work, and play; only now it's accelerating and compounding in unexpected ways. But just because we don't know exactly what form that tech will take (sharing homes on Airbnb or cars with Lyft and Uber for example) doesn't mean that the larger force at play (e.g., sharing) didn't have a certain predictability to it. It was almost an inherent -- and inevitable -- outcome of the very nature of the internet itself. And there are at least 12 such inevitable technological forces, shares author Kevin Kelly in his new book Inevitable. As we now move from an "internet of information" to an "internet of experiences" -- with virtual and augmented reality, AI-as-a-service, and more -- we need to accept the inevitable. Instead of fighting tech outcomes (things like tracking for example), we need to expect it, accept it, plan for it, and civilize it. It's not just about policy and laws, though (which should follow tech use); it's about new business opportunities (imagine if the music industry had bypassed its DRM phase!), cultural change, and new opportunities for humanity, too. Especially as the future of work changes. But productivity -- and even some forms of creativity -- is for the robots, argues Kelly in this episode of the a16z Podcast (where he is joined by a16z's Chris Dixon, Kyle Russell, and Sonal Chokshi). The irony is that while technology is inevitable, we humans are best suited for what is uncertain, inefficient, and full of failure. Machines may answer, but we will ask the questions. It's not just what we want, but what technology needs. Stay Updated:Find a16z on YouTube: YouTubeFind a16z on XFind a16z on LinkedInListen to the a16z Show on SpotifyListen to the a16z Show on Apple PodcastsFollow our host: https://twitter.com/eriktorenberg Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

The content here is for informational purposes only, should not be taken as legal business, tax, or investment advice, or be used to evaluate any investment or security and is not directed at any investors or potential investors in any A16Z fund. For more details, please see A16Z.com slash disclosures. Hi, everyone. Welcome to the A6 and Z podcast. I'm Sonal and I'm here today with our special guest, Kevin Kelly, who is the former founding editor of Wired. And he has a new book out called The Inevitable. Oh, I was going to say there's a subtitle, understanding the 12 technological forces that shape our future. We're here to talk about that because it's something we're all passionate about. And here with us to have that conversation, we have Chris Dixon and Kyle Russell. Welcome, guys. Great. Thanks for being here. Yeah, I'm so glad to be here.

So can you tell us about the book? Yeah, the book is trying to do a outline of the major trends in the digital tech arena in the next 20 or 30 years. So I think these kinds of long-term trends are fairly well set. They've been going for decades and they're going to continue. I'm not trying to talk about the specifics, which I think are unpredictable. So broadly, the telephone was inevitable. The iPhone was not. The internet was inevitable, but what kind of internet we had, whether it was commercial, private, international, or cosmopolitan was not inevitable.

and we had a lot of choice in that. So I'm talking about long-term trends, say, on the order like a Moore's law, where the specifics are unpredictable, but the direction is pretty clear. What's the point of saying it's inevitable? What's inevitable in the sense that I think we need to embrace them, rather than try to prohibit or stop them or lessen them, that we need to embrace. Imagine, you know, I don't know, 40 years ago, if you really believed Moore's law that that computers would become twice as fast, half as cheap every 18 months.

If you really believe that, if we're at society, you really accepted that, what we could have done with that knowledge, well, this is the way it's going to be. And so we could have altered our education politically. We could have accepted the eventual loss of certain kinds of manufacturing jobs. All these things would have been clear if we had accepted and embraced that rather than not believing it, resisting it, working against it. So is it, I can certainly see an implication for, for a country, like national policy, right, right, right, right. Like to embrace it and get ahead of it. And so you see countries like Singapore as an example, which has been, you know, very aggressive.

Or China today, for example, is very aggressively, you know, investing in education and computer science. And so I could see the economic angle. It sounds like there's also kind of a cultural angle. It's a sharing aspect, Uber, Airbnb. I mean, it makes no sense to try and outlaw those. or even to in some ways try to dismiss them as something that we can, other than that they're going to happen, whether it's those companies or not, that's the unpredictable part. But that aspect of huge decentralization, huge increasing the degrees and the dimensions of sharing,

those are all inevitable. And then also personally, when people are some of the other kinds of trends, copying, tracking, these are things that we have to kind of accept. So I would suggest that I can't see any counter force where there will be less tracking of our lives in 20 or 30 years. There's going to be more of it. We have to accept that. We have to work with that.

We can't stop the tracking. We can only civilize it. So it's better to accept that. And then, so for example, from a regulatory point of view, try to build smart regulations that assume, that Uber and Airbnb will happen and work with those trends as opposed to trying to fight the trends. Right, exactly. And also, I think there are technologies that can offset some of the difficulties or some of the harms that may be coming. They can also be offset with technologies, too, that can be invented and proposed. I think the job of law is to kind of put into place what has been worked out

in other places. And so I think, in... in a certain sense, the law is going to follow the technology and shouldn't really precede it. And so the legal thing can kind of solidify what is worked out on the street, so to speak. And I think smaller institutions, whether they be think tanks, I think we're seeing a lot of very innovative stuff happening in the civic arena with cities, particularly the big megacities. And so I would look to them in many examples of trying to figure us out. Sometimes the cities here are kind of backwards, you know, Paris outlawing Uber, etc. But oftentimes they're also very innovative as well.

Let's talk about the social behavior side of this because I almost question the premise that we have to, I mean, I agree with you. We have to accept the forces are inevitable. But it almost seems like there's a phase of fighting a technology that you emotionally, psychologically have to go through on your way to changing your behavior. I mean, on one hand, you're arguing that because it's inevitable, our behaviors will change. We accept that we walk around with smartphones and computers and tracking. That we wouldn't hypothetically have said we would accept it before.

But now that we do, is there a process where you almost have to go through it to get there? See, I'm a big believer that the way we steer technology is through engagement by use. I'm really, I find that most of the inventors don't even have any idea what the technology really ultimately be used for. I mean, Thomas Edison invented the phonograph, and we have his journals of what he was writing down what he thought this new ability to record sound was going to be. And his very first idea was that it would be used to record the dying, the last words of the dying. And then his second idea was we could do sermons. We can record sermons and distribute them. And he had a whole list of things.

And the very end, the very end was like, well, maybe we could do music. And he was the inventor of it. Right. So I just, I think it's only through use that we can find out what these things are. So I think I'm not talking about the fact that as soon as we hear something, we say, oh yeah, my gosh, I take it in. But I'm saying we have enough evidence of this stuff coming out and seeing how people use it to say, okay, let's listen to the technology. That's what I'm preaching. I say listen to the technology. The technology actually has built in often biases certain ways. ways that it wants to be used. So the internet is the largest copy machine in the world. By

nature, it's inherent in the thing. Anything that can be copied that touches it will be copied. So don't fight that from the good go. Don't fight that. Work with that. Work with the fact that copies are promiscuous and it's just going to go everywhere. This is the superconductor for copies. You have to, you can't battle against that. You have to say, okay, we can see how it is. You know, within the first four or five years, it was clear that this is the way it was going to be. Well, and practically, you don't have to go through a DRM phase every time. Like, you don't have to go through this ridiculous exercise. I mean, the original DRM inventors, they use the analogy of property rights management to come up with,

oh, this is how it's going to work on the Internet. And we could have just bypassed an entire phase. Exactly. I mean, can you imagine if the music industry had accepted? Right, from the front. From the front? I mean, it was been amazing. They would be, you know, they kind of just coming around to it now.

But, I mean, how far ahead they would have been if they said, okay, this is inevitable, this copy thing. We're just going to try to work with it. There's things to adjust, but let's accept it. So what are some of the things you think are inevitable? You mentioned earlier that 30 years ago, the Internet was inevitable, but the iPhone wasn't or the nature of the type of Internet wasn't. So can you give some examples of what now you think? There are a lot of them, but I kind of gather them into 12 ongoing verbs in the book. And one of them is a fancy word I called to cognify, to make something smarter, AI, basically.

So to me, AI is at this point inevitable. And it's been going on. There's recent acceleration because of GPUs and big data and the deep neural nets. So AI is coming. What kind of AI we have? What's the regulatory scheme? All the things, we aren't inevitable. We have a lot of choice in.

But the fact that things are going to be smarter and smarter every year, we have to deal with that. and want to civilize it, and we want to make the most of it, and we want to embrace it. So that's one example. Tracking, I mentioned, was another one. I called filtering. This is like that the abundance that we have generated

is so far beyond our tensions, Stephen, that we actually have to rely on filters of many levels and many kinds to actually deal with all this. We have this intermediated. So this is your Twitter, everything from your Twitter feed to... Everything.

Machine Curate. Right, right. And so there'll be more and more technological intermediates who are filtering for us on our behalf, against us, you know, the multi-sided markets, 500-sided markets, I don't know what it was. It's going to be very complicated. And I think managing our attention, I think that's something, again, that that's on the increase and will continue to increase. And, you know, there are downsides, filter bubbles. They're real, but we can work around them and overcome them. And all these are challenges and opportunities. You know, on the filtering one, let's talk about that for a couple of minutes because you have these examples of filtering by gatekeepers, intermediate, intermediates, curators, brands, government, cultural friends ourselves. Do you have any thought experiments that you've been, you know, thinking about what filtering looks like in this world that you're describing? There's a lot going on with our attention. So, I mean, Herbert Simon 50 years ago said, you know, age of abundance, the only scarcity with human attention. But it's really funny because I did these calculations about how much our attention is really worth economically. Like, it's pathetic.

The amount of the attention that you give to a TV, you get 20 cents an hour is what they're earning for your attention. But we're giving away the most precious thing, supposedly, in this economy. So why aren't we charging to watch an ad, to see an ad? Why don't we take Esther's Dyson's suggestion, you know, get paid to read email, right? I think this is one of the shifts that I can see coming, which is that we kind of flip things around, and you have a reverse market, so to speak, for attention where you're paying the high influencers directly, not going through the ad agencies. You're just giving them your advertising dollar directly to them to look at your ad, to be influenced or giving them the product, whatever it is.

And then secondly, you can have the consumers also generate kind of ads in the kind of a push system. So I think you could have a complete peer-to-peer ad system, the same kind of disruption that happened in the other industries, which is, you know, it's complete, user generated, you take out the ad agencies completely. In a way, isn't this almost the inevitable trend line of things like Yelp reviews where people build a personal reputation and like the way they think about themselves is as like, I am an experienced foodie and I, you know, my reviews matter. Yeah.

So sharing, just generally sharing, even though it's been going on for 20 years or more, I think we're still at the day one, day two of this, that's how far it's going to go. So there's another one of the forces share. Yeah, yeah, sure. So it's this idea that, you know, what can you share that isn't being shared right now, basically? And I think we're still at the very, very beginning of what we could imagine in that kind of collaborative way. And, you know, the general trend is this kind of decentralization that's been going on. But there's also this, the general trend of collaborating on another scale that we're,

we haven't been able to collaborate before. And most of the amazing, kind of miraculous things that we have have right now, like Wikipedia, are all about being able to collaborate on a scale and speed that would never have been possible before. This is one of my favorite things about GitHub. One of my favorite things isn't the developer community? It's when in the early days when countries like Germany put their entire code, their laws, and GitHub. And everyone essentially became an open source contributor. Right.

because they're now engaging with the law in a completely different form than they could have before instead of a static book. Yeah. So I think that that ability to collaborate and share at this sort of planetary, you know, a billion plus, whatever it is, that's new. That's big. That's still unexplored territory. That's still something that we don't have the tools, all the tools necessary. That's still something where it's way beyond in terms of what most people are even imagining. but I think that's what we're going to be doing the next 10, 20 to 30 years. Well, let's talk about interacting. Yeah, yeah. So interacting was the category that I was talking about, VR, among other things, but just this gesture using our bodies as password,

using our bodies as input and not just our bodies or voice, but getting away from the keyboard into full-body interaction with our computers. And I think there's a lot to say about. it. I was, you know, Magic Leap is very real, the visualization. And I think that what I became convinced by after seeing all the stuff, the void and the whole lens and all this, all the captures, is a couple things. One is that we're moving from an internet of information to the internet of experiences, right? So that's, that's the currency, its experiences. And it was this idea of knowing something not intellectually, but with your full body. You know, when you have these

that sense of presence it may be an artificial cartoon thing but it really is there and that sense of really being there it's not just in my friend of my mind it's much deeper and that kind of deep sense of experience I think is very powerful

and very different than the kind of information stuff that we get the world of Wikipedia and PDFs and pictures in that sense of oh I feel that that's something that I, and I think that moves and makes... Yeah, no, I think that's right. I think, well, I would say two things about VR. One is to talk about inevitable.

Like, to me, it feels like the path, like some of the current systems have problems. They're too expensive. They're, you know, the resolution isn't what we want to be, the, you know, the hand tracking, field of view. But it's very, very straightforward path to... So I always say to Kyle, like, we think it's like 1977. This is the Apple two. Right. Like, the Oculus is the Apple two.

and if you go back and look at it, it wasn't until 81, the PC came out that it really kind of exploded. So we're a couple years away, but it's very, very clear what you need to do. We need to increase the resolution. We need to, you know, all sorts of display technology which magically and others have hand tracking, like deeper immersion,

foveated rendering, you know, so to like take... What's foveated rendering? That's a way to basically get the effect of much higher resolution without having to have GPUs that are... Yeah, yeah. Basically, it's

you do very high-res where the eye is looking and lower-res around it. That's actually how your eye actually sees things, is that you see, like I'm looking at you now, I see you in high-res and everything else in low-res, and so render the same way, which makes you get much more performance out of existing GPU. So anyway, it's a long way of saying. It's good enough to improve is how I say. Yeah, yeah, it's good enough to improve, but it's like very clear. And in a way that's different than, in my mind, the Internet in 93,

where you had a much more complicated kind of network. network effects. Inter, like, you needed, like, web apps to exist, which reinforce the infrastructure. Like, here, it's just kind of a straightforward. Right. Hardware problem. Like, right, like, and we know how to do that. And people are working on it.

A lot of money is going into it. So it's going to happen. Right. And, and then, yeah, then the question is what happens when you're in there. Like, I think that VR will be the most social of all the social media ever. That is so counterintuitive to me. I feel like it's a much more intimate experience than Google Hangouts or something. Like, you know, even though it's symbolic.

Because it's like, in your brain, you're like, yeah, Kyle looks kind of symbolic. but I hear his voice perfectly, his body moves perfectly, and your brain kind of fills in the rest. Actually, the two things, having the voice, having the body language from the person, and eye contact. That's right. Those three things are probably all you need to have that presence. And in fact, they're more, and what surprised me in experiencing it,

and this is your point about having to use it. It surprised me that was more important than seeing the actual pixels of their face. Exactly, right, right. And what's so funny is, even with the eye tracking, Like within the next generation or two of headsets, we'll probably have cameras on the inside of the headset, one for foviated rendering, but also just to track your eyes so that when I make eye contact with you in the virtual experience, we actually make eye contact. I think to me that really interesting tipping point will be at what point do you no longer need to fly across the country to, you know, all the people, all the business people and the suits with the Windows laptops or whatever, you know. The reality is today, like, why do they fly? Because to sell a million dollars deal, you still got to look someone in the eye and you need that level of intimacy.

And that's how business works. way. But that's going to, and also, and then like group meetings. Oh, yeah, yeah, yeah. You still need to be in the office. But at some point, that's going to go away pretty soon, I think it's good enough. I think, yeah, I mean, I saw the uncorporeal capture, 3D volumetric capture, and it was stunning. And it was so real. Basically, it's a, it's a 3D capture of a person moving. So it's not a still image. It's a moving person. And you can walk around and inspect them from any viewpoint. It's the most realistic floating hologram you've ever seen. It is. And it's so.

And the sense of presence was so real to me that actually it was uncomfortable invading their space, getting too close to them. And a similar company, ATI. One of the demos they have is a woman in a bikini. And the point is that it makes you uncomfortable. Like, I tried the demo and I found myself looking away because it felt like she couldn't see me and I felt weird. It was almost foyeristic. Yeah, AI was also very similar to that. And they had the very moving thing where the mother is recapturing her child's birth. And you can. you kind of review it. And I think it is sufficient. And I think it's very close to becoming something that would be substitutable. And by the way, that's something I'd pay a lot of money for.

If you had real teleconferencing that had a presence. So looking at, again, these through lines of things that are inevitable going around the specifics. So looking at a category like VR, which parts, like let's say VR fundamentally, looking at its name is tricking your very, very, aspects of your brain into thinking what I'm seeing is real, what I'm hearing is real, what I touch is real? What's inevitable about VR? Is it that we're going to all have hooks into our brains that we can directly trick every aspect of the experience of reality? What I'm saying, what I would say is inevitable is just a direction. Okay. The direction, what's inevitable is every

year will have something that's sort of more, more realistic. And what we don't know, what we don't appreciate it right now is the degree to which, you know, what we're seeing. Reality is complex and the number of different factors that we use to discern whether it's real or not. And so there's a, so that list will continue to grow, even as we become better and better at it, we're also discovered that there's, that list is longer and longer. So it's not like we're going to reach the destiny. It's you're, we're on a progression, a trajectory, which is every year we'll do, we'll, We'll tick off another box and invent two new boxes to tick off.

So it's a constantly receding goal, so to speak. It's like AI. AI is basically defined as something we can't do yet, because as soon as we do it, it's machine learning. And so VR is going to be the similar thing, where it's just a direction that we're going in. I think the most surprising thing for me about when you talk about, when all of you guys are talking about VR,

I always sort of heard that it would never be social. And so I thought we'd interact more with objects than we would other people. That was very counterintuitive to me that you were actually all talking about a social interaction. Same thing that happened with the Internet, though, too. Exactly. That was the accusation, you know, teenagers in their basement being asocial. Right. And that's exactly the same image.

All of the VR people, their guys, you know, in the room. But no, no, it's going to become so social. And it's because we find people more interesting than objects. I saw the second life and high fidelity's second. Life, the Sansa Project in VR. And I was, it was a, it's not second life. It's something else. Second Life was kind of clunky. It was rigid, it was robotic. But once you have the avatars being driven and reflecting the body language of the person and having eye contact and having that kind of sense of presence, you suddenly could imagine spending hours hanging out in these worlds

sharing experiences. It's mind-blowing. I mean, but the example Kyle shared of the company having this woman in a bikini and using this real-world interaction of, whoa, I can't be voyeurist rick. I have to be respectful. There's also an element, you talk about this paradox where you can be real and you can also be very unreal because you can manipulate things in ways you can't do in the physical world. So what happens? I mean, have we seen anything yet socially here in terms of new unexpected social behaviors that are emerging? Is it too early?

Yeah. No, you're right. to look at that. I don't think we've seen enough street use of this yet, but we will. I mean, one can kind of imagine what would trolling look like? What's the VR? Exactly. What's VR trolling going to look like?

Well, the question that I was asking all the people I interviewed was, what's the VR equivalent of lolcats? Oh, really? Because that's what's going to, most of the stuff in VR will be low cats. Well, we don't have to have cats. We can have dragons and dinosaurs. That's what it is. But there'll be some. minimal, viable, you know, active VR experience, whatever that is, it'll be a low-cat equivalent

and that'll fill up most of the hardest. And so what is that? We haven't talked about deep learning yet. And AI, I mean, you talk about cognifying. That's a word you give it, which I actually don't know how much I'm crazy about that word for the record. I mean, to be honest. It's a verb. We don't have a good verb to make something smarter.

Right. So do we want to say smartifying? Smartify. What is that word that we say for? Amified. Oh, God. Yeah, exactly. Smartify.

So making things smarter, making things intelligent. What is that verb? So I use cognify, cognification. Yeah, it's a clunky word, but I think it's actually an accurate word. I want to know why you, I mean, everyone says this time is different. But you know what? People said that the last fucking three times.

Exactly. And so why is it different? Why is it different this time? I think, you know, the neuronets from the 50, 60, 70s, and 80s, they were just hampered because they didn't have the right. The technology was they were ahead. They needed to have the parallel processing not on the supercomputer but the cheap GPUs. So they made affordable.

They needed to have the big data that they didn't have to do the training sets. They needed millions and millions of examples, not just thousands. And I think, you know, the deep learning neural net hierarchy that the Canadians kind of invented was necessary to kind of really do the big neural nets. You just couldn't do them in one big. The proofs in the pudding, right? Like the, if you look at the results. image net is a good example where it's been the error rates for 30% or something and now are better than humans.

Right, right. So for me, it was a perfect storm with these three technologies converging finally. And now you had something where you could have parallel processing really, really cheap, big days. And so now it can kind of really go fast. And now that there's proven that we have advances, then the money will flow in and it'll just keep going. So I think, I mean, we could hit another stumbling block because... You mentioned money flow. That's the other important thing is there's a business model for this now.

Exactly. There's very high stakes involved. Right, exactly. So that also helped. You don't need government funding now. You've got corporate funding. And so my hypothesis is that AI becomes a commodity on the grid and that it's served like electricity. You purchase it like you purchase electricity to do what he wants.

So you take X and you add AI or you take AI and you add X and that is sort of the great big frontier right in front of us. Where like the industrial revolution, the farmers making all these cool electric gadgets didn't have to generate the electricity. They were just purchasing it. I think just kind of this equivalent of AI as a service is one, it's a business model for some of the generators. And two, it just makes it, you know, it's the back end. It's invisible. I was like to say it's invisible. That's the most compelling idea.

Right. It reminds me of Brian Arthur's second economy. I recommend anyone listening to this podcast, highly recommend that they read this article. He wrote called The Second Economy, where he talks about this invisible network of, it's like the trees, the Pando trees. And it's entirely beneath the surface. Exactly. So in 1920s, Sears Robot catalog, mail order catalog, had the big home mode. that was going to power everybody's home.

And it was about the size of a microwave oven today. It was this big, massive motor that was going to be in your home, and then it was going to power all the things in your home. Well, what happened? It succeeded because those motors became invisible. We have hundreds of motors that we don't even see in our home, and that's because they succeeded by becoming invisible. And that's sort of, I think the AI is going to become invisible.

And the second thing I would say by AI is that the whole purpose of it is we want AIs to think differently than humans. So we're going to make many hundreds, thousands of species of thinking. And that's the purpose of it is to have more than one kind of thinking and to work with that kind of thinking. So people will be paid by how well they work with the AI. And, you know, the Kasparov, he blew when he lost the chess championship to a machine. He said, you know, if I had a...

access to the same database of every move and chess ever was, I would have won. So he made a new league, the free chess league where you could play with computers or by yourself. And those combination of a human plus an AI is called a centaur. And in recent years, every single world chess champion has been not a human, not an AI, but a human plus an AI. And that's what the goat is going to go in the same direction. It'll be the best go player will not be AlphaGo. It will not be Lisa Dole. It will be Lisa Dole plus AlphaGo. Well, I mean, to

take that to maybe a more relatable example, the best drivers on the road today, besides maybe someone with Tesla, would be everyone who has ways open on their phone. Exactly, right. And Watson and going to the diagnosis, I think correctly, pitching it is something that doctors

were used. So the best diagnosis would not be the best doctor, not to be the best AOLB. It would be a doctor plus AI. And so this idea of because AIs will think differently than us. And so we have some scientific problems in quantum gravity, dark energy, whatever it is, that are probably too difficult for us to solve ourselves or their own kind of thinking. And what we're going to do is invent other kinds of minds of AIs to together solve these problems.

Did you watch the AlphaGo YouTube video? So there was one moment. I forgot it was like Game three or four. Game three. Yeah, exactly. There's this move where if you're watching the announcers, they all like gasped. Yes, right. the computer to this move and they're like, that must be a mistake. Of course, it's a mistake. Because it would be like, I don't play Go, but I play chess. It would be like moving your, like sacrificing your queen or something, right? Like some just terrible move. And they all were like, there's an error in the system and it turned out later on to be a brilliant move. Right. But it was a move that apparently, I don't know, go as well, but no human would ever

made that move. And it was like literally gas inducing how bad it seemed, which shows you the difference in the way of thinking, right? But what was also wonderful is that Lisa Dull said it was beautiful. He said it was beautiful. And so that's an important thing, is that I think we overestimate our own creativity in a sense. I think we're going to look back and understand a lot of our creativity is very mechanical. Oh, I completely agree with you about this.

And so, yeah, I think definitely AIs will be creative. They definitely will do creative things because creativity is going to be like driving. It's like, oh, of course, of course machines can be creative. Once it happens, we'll all recognize the fact that our own minds, our own intelligences are just really a suite of hundreds of different types of thinking. So what we don't have, we have a lot of perception right now, the AI, but we don't have still symbolic reasoning, deduction, and those engines still have to be built. And they're also difficult because unlike the type of learning that most deep learning works on right now, you require big data set. Right. And actually humans, they did, I was looking at some studies for humans, babies learn to identify

between cats and dogs in only 12 examples. That's right. That's exactly the, I was actually thinking of development psychology. This field is moving so quickly. And if you follow, there's a whole, like a bunch of papers came out recently around this, what they call one-shot learning, which is exactly this, the idea of sort of algorithms that use fewer. The other kind of holy grail and AI right now is to unlock, to be able to use unsupervised learning

data sets. Right, right. So, you know, like today you basically need, the only thing. you train on is I take a set of, you know, whatever sentences, and then I go to Mechanical Turk, and I label them, and I input them to my system, and I train it, right? Whereas if you could take every sentence uttered into Siri and, which is unsupervised data and use that, you suddenly unlock the 99.9% of data in the world's unsupervised, which at all seen, I mean, like, we'll see, like, I'm drawing, I'm drawing lines through dots here, but like, it seems like all these

kind of obstacles are tipping, you know, are sort of falling over right now and that they're making such rapid progress. I think what humans are going to be good for in the next 10 of 20 years is these interpersonal roles like, you know, nursing at home, coaching experiences, again, coming back to experiences, and asking questions. And so I think if you want to answer, you're going to ask a machine. If you want to question, guide us through the uncertainty. Humans are really going to be really good.

Because I think machines don't, right now currently work well with uncertainty and fuzz. And I think that's one thing that we're going to be good for for a while. What do you think of the people that think is dystopian, that it will take all the jobs, it will, you know, Skynet? Well, there's two separate things. The Skynet thing, I think, is a cheap Hollywood trope that is highly, highly unlikely. The taking all the jobs, I think that jobs are made up a bunch of tasks. a lot of the tasks that we do will be done by bots, and therefore they'll change our jobs.

This is the idea of working with these things. So, yeah, so there'll be a lot of tasks that are going to be replaced. But I think that changes the jobs more than replaces the jobs. And who was the economists who said human needs and wants are infinite, you know? So, I mean, as you sort of move up the stack, people will want new, whatever, I guess we're all become filmmakers or something here. Exactly, which is what we want, right? So the other way I say is any job that can be specified or can be specified in terms of efficiency or productivity is a job that humans should not be doing.

And it will be a job that would go to the bots. Productivity, by the way, which is the measurement of GDP, productivity is for robots. And what humans are really good at are all the stuff that are inefficient, like science and innovation where there's high rates of failure. That's inefficient. and in expression and interpersonal relations, those aren't efficient or productive. And so all these things are left. And it's not just cerebral work.

I mean, it's not like you have to be white color. I think interpersonal stuff, coaching, nursing, those are things that you don't need to have a degree to do that also become really very valuable. And by the way, those are the only things that are increasing in price over time. everything else is dropping in price. And so that's where our money's going to go.

That's where our money goes. That's where the income is. So I'm totally optimistic that we'll have more jobs, more opportunities for all kinds of people. I think one reason people are pessimistic in general about technological progress is there's an asymmetry in that it's very easy to imagine the jobs that are destroyed and very hard to imagine the ones that are created. Exactly. Ten years ago you wouldn't have said, oh, social media. manager and, you know, I don't know what, like, you know, I don't know, machine learning the trainer or whatever, all these new jobs we have, like you would never have imagined them, but you

could easily imagine the opposite, right? Yeah, so you travel back 150 years to the 70% of Americans who are farmers and you tell them, hey, all of you're going to lose your jobs. And they would say, what would we do? Oh, yeah, web designer. You know, you're going to be a yoga coach and something. What are you talking about? There was, there was, maybe you tweeted as a graph. It was the drop in typographers and then the corresponding rise in graphic designers. And there were, of course, more and better jobs as graphic designers, which you never would have predicted. I mean, that Photoshop would be sitting on 10 million desks and, you know, back in the typography era. Much more controversial, and not too many people believe this, but I believe that, like, you know, everybody's photographer now,

that on average the photographs taken today are 100 times better than the photographs taken, you know, when photography was first invented by the best photographers. We are not only, so we're better topographers now than we were before. We're better photographers now than we were before. It's not just that there's more of it. It's actually better, too. I know this because I started off as a photographer. And I tell my kids the story because I went to Asia instead of going to college and I had a backpack with 500 rolls of film.

And I was taking two roles a day in Asia, which is there were 36 exposure. So that was 70 pictures a day. and I would come back and tell people what I'm doing, and they would tell them I took 70 pictures a day, and they were just completely mind boggled out of, they said, how is it possible? Because their brownie camera had a roll of 24 exposures, and they would send it off to Kodak and come back,

and it would have pictures from Christmas and Easter and Halloween on it. They would do like 24 pictures a year. So 70 pictures a day was considered insane. Well, the only way you become good as you're doing a lot of it, It's the fact that we now have, you know, people pay a lot more attention. They look at them. They try and do it themselves. Rapidly iterate.

You can waste. I mean, you can now take 200 pictures of one baby. Right. You can waste all that. It's so important. I agree. We take it for granted that this is that kind of abundance to design for a world where you can waste more than transistors is pretty freaking incredible.

Right. I mean, I was early enough to use the dialogue search system. This was the very first search, like before Google, and you had to pay, I don't know, $20 a minute or $60 an hour. I forget what it was, to do a search. And you would actually have to plan out all your searches beforehand. You'd write them out. Oh, my God. And here's what happened was you would never waste a search.

And the genius of all the stuff that we've come is because you could have searches that you could waste on. You could search for your own name. Well, now you can go down like these little rabbit holes. You start searching for one thing. You end up like you starting Kim Kardashian. And people and what used to be like computer language like bullying expressions and things like this, everyone's now incredibly proficient because they spend all day iterating in searches. So for me, a big tipping point, a big tell is where can we waste things that we couldn't waste before?

And so AI is going to be very close to that where you can basically take the world's best AI and completely waste it in ways that you couldn't before because it was. It's too precious. In fact, it's built and waste. And that's when all the new things, innovations, and all the new wealth comes is when you actually have this resource that you can waste for the first time. No, and this is where people say, you know, robots will take our jobs and we'll, to deal with that, people will have to reinvent themselves over and over. You won't build to have this, like, static base of knowledge that you got in college

and that lasts you your career. But AI, you'll be able to say, like, tell me how to do this new thing. Just knowledge, experience itself will be just like run like water. I mean, it's cliche as it is. It's like, I know kung fu. Yeah. Right, right. Well, there was somebody who did kind of an art project where they had a robot sorting rocks.

It was a waste of time. But that is the genius, which were discovered, is when you can have robots wasting their labor. That's the genius when the new things will be discovered. Well, I feel like the title of your book should be the inevitability of waste in a good way. Also looking forward, you know, I think here in Silicon Valley, what we assume to be inevitable is a combination of things we've seen from sci-fi and the intercept of that with what companies to, today are good at. So we assume AI is inevitable because Google is really far ahead compared to companies and we think, oh, deep neural nets are advancing very quickly.

What's inevitable that maybe we haven't quite identified in the zeitgeist here? That's really good question because there was actually a data list at one point of there's two kinds of expectations or inventions. There's ones that were expected and then unexpected. And so like, you know, flying machines were kind of expected. but nuclear power was not. There was no science fiction before nuclear power about nuclear power, where you could take a little bit of matter and it has all this energy in it.

That was not in anybody's. And so I think AI and VR have both overly expected. I mean, they've been expected for such a long time that it's almost escaping from that. It's really hard. So the kind of ones that aren't expected, well, the internet was actually not expected. There's almost no science fiction stories about something like the internet.

So that was an unexpected technology. So the example to look at would be like something like Heinlin, where we knew that we're going to be traveling to planets super far off. That won't be a problem. But how would we communicate across such a large existence of humanity? It was almost, yeah, it was not very, not very expected. So you're asking like if there's ideas. I don't talk about this in the book.

This is separate from that. But I think, did you read the Nexus trilogy? Oh, I want to read that. Ramiznams. Right. So he talks about this kind of telekinetic network where you kind of like telekinesis and you're kind of mind melding. And I think he does a brilliant job, a really kind of imagining this thing that I would say, well, now he's trying to predict it. But, I mean, in generally, there hasn't been a lot of expectations about that.

And so he's taking that. I kind of feel like Twitter is like a proto-hot. hive mine. So I kind of think that's inevitable too, though. One thing I think if you go back and you look at the predictions in the past, one thing they tend to do is they take the field that was really successful the last 30 years and they assume it will be the same field.

So for example, if you look at the steampunk stuff, what were they really doing? They were taking mechanical engineering, which was the big success of last century in the Industrial Revolution and they sort of extrapolate forward and say, okay, we'll have mechanical engineering computers. Yeah, yeah. And I wonder sometimes if we're doing that with computer science today. And that

actually, like, so what we're now imagining, like, I'm optimistic about these trends you're talking about, but I wonder if they'll look back and they thought, ha, ha, in 2016, they thought computer science would be the story when actually it was biology, you know, and so I wonder if we're making that same mistake where we take the thing that's the thing that's worked so well for 30 years. Now, you know, it could be both. It could be more complicated. I think it is both. And, you know, as somebody in the last century in 1994, I was writing a book about the neobiological civilization, I think biotech in the long time. term will have that kind of immense power and cultural power as well. But I don't think in the 25-year horizon that I'm talking about. So I think beyond that, yes, I think it's wide open, but I think, you know, for the next 25 years, I think this is what I'm talking about. I think computer science will continue to run everything. Software will continue to eat the world and so forth. But beyond that, I think it's way wide open and certainly biotech may eclipse the software.

You know, software. Well, everyone, thank you. Thanks, Kevin, for joining the A6 and Z podcast. Thanks, Kevin.