Mark Bidwell 0:38
Hi, this is Mark Bidwell, welcome back to the OutsideVoices Podcast. This episode, this week, is a little bit different from previous ones, it’s much shorter. The reason being is that I recently interviewed Kevin Kelly, a past guest on the podcast. The previous episode was called Failing Forward: The Formula for the Next 10,000 Startups, which was based on his book The Inevitable, which mapped out the key technology drivers shaping the world. Anyway, he’s got a new book coming out called Excellent Advice for Living. And while we wait for the release date of that, which I think is in a couple of months, the podcast has been embargoed, but in the meantime, because Kevin has an amount of such diverse interests, part of the interview we talked about AI, which is one of the areas in which he is genuinely a world leader. And given the interest in the media, and in boardrooms around the world, around the topic of AI, stimulated by the exponential uptake of ChatGPT, and the subsequent evolution of chatbots and AI, I felt it’s worth sharing with you some of his insights into the topic. So this is a very short interview. But, unless you’ve been living in a cave or under a rock, probably like me you’re very interested in how this technology is going to shape out, whether it is just one more iteration of the Silicon Valley Hype Machine, which we saw with Bitcoin, crypto, we saw with web 3.0, which fizzled into nothing, or whether this is something far more fundamental, which actually threatens a number of our global institutions by relating to democracy, relating to the economy, relating to what it means to be human. So I wanted to share Kevin’s thoughts on AI with you in this episode. Check out the previous episode interview with Kevin, as I said, it’s called Failing Forward: The Formula for the Next 10,000 Startups, the link will be in the show notes. And if you’re interested to learn more about this topic of AI, I’d point you towards a couple of things; I’d point you towards the Lex Fridman Podcast with Sam Altman, the founder of open AI, which is fascinating, very provocative, very stimulating. I really enjoy Lex’s podcast, he’s a really humble guy, but he works very hard to bring out the best in his guests. And I think Sam Altman, I know he’s quite a polarizing character, but he feels like he’s in it for the right reasons, although, by his own admission, he’s deeply nervous about the implications of what his company has created. And the other place I’d look to is Ben Thompson Stratechery, I’m never quite sure how to pronounce it, nor is he actually, but it’s as a podcast well worth listening to, and it’s actually worth subscribing for 120 bucks a year to his articles and his podcasts, because he really does nail new technology and bring some pretty fresh insights to the impact that technology has on business. So I’ll put those two links in the show notes as well. Hopefully, you’ll enjoy this very brief episode, which is a back end of my conversation with Kevin Kelly. And as soon as the book is out, listen to the full interview because he is a remarkable guy, I’ve learned a huge amount from him. He’s the father of the idea of 1000 true fans, which kicked off many hundreds of 1000s of creators’ activities on the internet. And he really understands technology and the impact it has on humanity better than many, given the perspectives that he brings, and the fact that he’s been active in the space for well over 40 years. I’ll leave it here, here’s Kevin Kelly.
Kevin Kelly 4:08
I’ve been putting out challenges to people to actually name, give me a real person who’s actually lost their job to AI. And there hasn’t been any. I think I might be able to find one person in the transcription business, so the business of humans taking audible sound and transcribing it. If there were people who actually had their job that way, there probably aren’t very many, because that is definitely something that the bots can do now, today, very well. But in general, what we’re not seeing is this large scale unemployment due to AI. At the other end, what we are seeing and what I’m doing every day now, is using these things as interns. That’s my current frame of reference, my model. CahtGTP is an incredible intern, is an intern in the sense that you offload all kinds of things to it, but you have to check their work and you don’t release their work as your work, you have to kind of work on it. It’s like a first draft intern doing the summary, they’re doing all this stuff of behind the scenes and working with you, for you to present something. So even in the AI image generators is very rare when the very first thing you get is what you can use. I mean, you can be surprised by it, but it’s very hard to get it to obey where you want to go. And so there’s this long conversation of AI whispering, incantations and working with them getting better at it as if they were an intern. Okay, yeah, here, that’s good. Go back and do this. How about that? Okay, let me know, a little bit more of this, and so. And so this idea of partners, I think is being borne out. And what we’re seeing is that, say, in the image generators, there are some people who are really, really good at it, and you look at it, you say, how did you do that? And it’s sort of like Picasso says, well, yeah, it only took me 10,000 hours of practicing with this and doing it over again, and having seeds, and having my special prompts that I use, and knowing how it works. And so, until there are going to be people who are going to be better at working with the AIs than others, and also, they have personalities, there’s making the images, Midjourney has been engineered and trained to be more arty. DALL E-2 has been engineered and trained on more photographic type stuff, Stable Diffusion has its own kind of personality and niche and biases. And so, we’re going to use different tools for different things. And some people are going to be more comfortable with certain kinds of personal AIs, and others will find it too alien and don’t want to deal with it. And so first of all, we’re gonna have many of them. They’re not one AI, that’s coming true. And I think, the idea of having partners that we’re going to mix in our own human preferences into these partners. And so there, there’s going to be people who just can’t deal with the alienness of certain things, but are going to be comfortable with other things. And so, in people trying this, we’re going to have to understand that the first couple of API’s you try may not work for you, and you should not give up, but try a different one.
Mark Bidwell 7:14
I presume you’ve been following what’s happened recently about the jailbreaking in Sydney, and these are extraordinary stories. The second point, and I’ll come back, the reason I raised this is the second comment you made five years ago, when we spoke, was that you were concerned about, you were worried that we were going to abuse AI and treat it like slaves. And they were trying to jailbreak this, the Microsoft.
Kevin Kelly 7:42
Right. These are black mirrors in some way that they reflect how you treat them. And the full dialogue of this is yeah, I mean, these are imaginary friends, right? So the fact that you can get an AI to say something doesn’t mean the AI believes that, it’s like, they’re just mirroring you. They’re mimicking humans, and they can exhibit many personalities, because they don’t really have a unified personality. And so these are the early days, here’s what I would say about this. This is maybe where we are right now, which is that these versions of AIs are auto-complete, they’re mimicking humans, it looked at all the things that humans have drawn, all the humans, everything that we’ve written, and they’re gonna say, based on everything that you’ve written, or said or pictured, I’m going to predict that this is the next thing. You’ll start me going, and I’m going to complete it mimicking humans. So they’re mimicking the average human behavior, which is not necessarily the noblest human behavior. Alright, so on average, humans are probably racist, and sexist, and everything else flawed. And so these things are going to be like that. And here’s the important thing, we are going to demand that these bots be better than us, and we’re not going to accept them to be like us, we want them to be better than us. And that’s actually something we can do. We can actually program in ethics and morality into these bots. But we don’t have any idea what it means, or how to be better than us. We don’t have any consensus on what they would mean. Our own ethics and morality are so flimsy and shallow and inconsistent, that we don’t know how to make them better. And that’s the conversation we’re gonna be having is okay,we don’t want to be like the typical human, we want to be better. So what does that look like? How do you behave better? What is the consensus on that? And we don’t have any consensus on that right now. And so, that is the thrilling adventure that I think we’re headed into right now, is trying to elevate the AIs so that they’re better than us, even though we don’t know what that means.
Mark Bidwell 9:56
Yeah, yeah. I mean, there’s something very compelling, I haven’t been engaged with them in the way that some of the others have been, there’s something very compelling about being able to engage at the level that some people are reporting. And for them to jointly go on a journey of self-discovery with a co-pilot or an intern, it’s actually raising your game or giving you insight. So they’re not hallucinogenic insights, but actually can leave you as a better person. It’s quite exciting.
Kevin Kelly 10:24
It is. It’s very exciting. And there’s another frontier that’s epistemological, about how we know what’s true, and how we accept and trust things, and how we decided that we know things. That’s another frontier that these are unleashing, they put a crack into the world of the ways that we accept truth or not, we realize oh, there’s a whole bunch of assumptions that we can no longer rest on, that we have to actually become more precise about. And that’s the second frontier that we’re going to be headed into, which is trying to automate these trustworthy.
Mark Bidwell 11:02
Yeah, and I mean, you’re clearly optimistic about it. You wrote a wonderful piece a couple of years ago about the case for optimism. I’m assuming that’s still intact for you? And if so, are there any things in particular, that you are particularly optimistic about that you can share? And also the counter as well, any high points or low points in your view of optimism?
Kevin Kelly 11:26
I am optimistic, and I am more optimistic than I was, I think I have a natural sunny disposition. But I have actually been crafting a deliberate engineered optimism, which I think is necessary today. And I think hindsight would prove that most of the things that we’ve accomplished had been done by optimists who believed it was possible to do, so yes, I am optimistic. Just recently, this past week, I got to visit some Silicon Valley startups that were doing NeuroLink, that were doing the brain computer interface, implanting things SO the brain could control a computer. So you would think, and the computer would understand what you’re thinking, which I had thought was 100 years away. And it turns out that no, this is actually something very close. And there’s more than one company doing it, there’s Synechron, there’s OpenSea, and they all have very different technologies to do it. They’re very complicated, but they’re actually starting human trials. And I saw the monkeys doing it and it’s amazing. So in terms of optimism, and a sense of something that I didn’t think was really feasible, I thought it was really science fiction-y. I didn’t think it was near-term feasible, and it was possible and inevitable, but I thought it was, again, not in my lifetime. And now, wow, there’s something happening there. So that’s optimistic. I’m optimistic about green meat, lab grown animal cell meat, as someone who doesn’t eat mammals, I’m looking forward to that very, very, very much. Where that’s, again, you can make meat, you can have animal cells existing, but you also can make meat that’s better than the meat that we have right now. That’s maybe tastes more meaty, or the taste is even more delicious and still being of animal cells. And so I’m optimistic about that. But there’s lots of things to be optimistic about.