Business, Innovation, and Managing Life (July 31, 2024)

Hello, everyone. Welcome to another episode of Q and A about business, innovation and managing life. And I see a bunch of questions here saved up. So. So let's take a look at some of these. Oh, here's a. Here's sort of an easy one from Yanao. What can you tell us about the next Wolfram Language release? What are you most excited to see added to the language? Well, actually, we're planning on this happening later today, so tomorrow I will do a live stream about all sorts of cool things in our version 14.1. It's kind of a shocking fact that it is the 36th release of what's now Wolfram Language, what we used to call primarily Mathematica, but it's the same strand, the same technology stack. The first version came out June 23, 1988, and as I say, we're now on version 14.1, but our 0.1 versions take tend to be pretty big. Basically, the strategy is we have a lot of ongoing research and development and we are sort of nowadays chopping it up. Twice a year we release a new version. And so whatever is sort of in the R and D pipeline at that point, that's the version. And we try to make it so that the integer versions kind of wrap up large domains as sort of finished things, and the 0.1 versions are sort of the incremental pieces of those domains or the parts of those domains as they're getting built. But in fact, in some ways the 0.1 releases can actually be bigger than the integer releases because they have kind of the main structures getting built towards sort of finishing off some domain, and the integer releases end up being kind of just finishing, just putting the final coat of paint on those domains, so to speak. But it's sort of interesting if you look at what happens in successive versions of our product, it's kind of a mixture of sort of incremental polishing of things that were already there. It's amazing to me that when it comes to, for example, user experience, we're still inventing things that make the user experience smoother. Even though we invented the idea of notebooks in 1987, very long time ago, we're still smoothing the user experience and discovering we just invented something yesterday. Actually, that I think is a really cool idea that makes it easier to sort of type correct Wolfram language code in notebooks. It's an idea that we could have had 20 years ago, although maybe some of the scaffolding for it wasn't there then. But any case. So, you know, part of what we're doing is polishing the capabilities, making things more efficient. And part of it is, is going off building new things, both things that kind of build around things we already have and sort of spurs out into completely new areas. So a lot of different things coming together in new version and I suppose the repeated story of my life, at least I can sort of measure my life in versions of Wolfram language and Mathematica. Although I have to say that our teams nowadays have a pretty good system going for the actual process of releasing these things. But what you'll see happening is over the year or two before release, there'll be lots of design reviews, most of which we live stream kind of developing, figuring out how these pieces of functionality should work, then they're implemented and then they're deployed in versions that come out. So you can typically see sort of what's coming for a year or two, sometimes more than that in the live streams that we're doing as we kind of develop these pieces to be added to the product. Let's see, very different question from anon here. Do you worry about the increasing appearance of incompetence in the world? Well, let's see, it's Yes, I do worry about it, but I'm not quite sure whether it's one of these things where at any given time in history there'll always one will always be saying everything is going to hell, so to speak. I think it's a little different now. I think part of what's happened is that there's an awful lot of experts in lots of different areas. And one of the things that's happened is the kind of expectation of experts, expertness as things get sort of made into these narrower and narrower silos and people are like, well, I'm expert in this, but if there's something that happens that's just a little bit away from the this, it's like, well, I don't know anything about that. Or more to the point, I'm not really capable of thinking about the thing that's outside of my specific expert silo, so to speak. You know, that would be, I suppose, one diagnosis. I think the other thing that can happen is when it comes to sort of education and the expectations of education, there's a certain sense in which a lot of fields have been kind of made procedural and you're kind of, you know what you learn, okay, you're going to learn, let's say, you're going to do something which I consider fairly non procedural, like run a company. Well, in Principle, you can go to school in sort of how to run a company or how to start a company. And in principle, people will claim, well, there's just a procedure for doing it. And you just do steps 1, 2, 3, 4, 5, and so on. And so long as you know those steps and you've kind of memorized aspects of those steps or you've done exercises around those steps, it's all going to work. Well, I think what often happens is that isn't true, that by setting things up in that very procedural way, you're building something that is based on basically very brittle, where if things actually work exactly the way you expect them to work, then, yes, you can just follow the steps and all will be good. But as soon as something different happens and you didn't kind of know the things outside of that narrow direction that you learned in, then you're kind of lost. And I suspect that there's some of that that's going on and the kind of. And perhaps because there was sort of, in a sense, perhaps people perceived there being less to know in the past, there was perhaps a tendency that, you know, if you were going to be educated, you would learn a bunch of different things. And that was what happened. Whereas now it's like, well, there's so much to know. You've got to specialize in this particular area. You know, I do have to say that with the kind of rise of kind of computation, AI and so on, I think the value of, of sort of deep, specialized knowledge is perhaps on its own is perhaps being decreased because it's kind of like, well, you can just go, you know, go compute it, Go look it up. Go get the AI to help you look it up, whatever else it is, and knowing it cold is perhaps less important as such. But I have to say that at least story of my life and efforts is that knowing a bunch of things with reasonable clarity is super useful for figuring out other new things. That is the idea that, well, you don't really have to know anything is a big mistake, but to know just one thing incredibly well and nothing else also seems like a big mistake. And I kind of wonder. I mean, I suppose it's a very common thing in kind of human factors discussions about sort of how mistakes get made. I think it's usually called fixation error, where people just sort of say, we're just fixating on this one issue. We got to solve this one issue. But there's a giant freight train barreling down that they don't notice. I think a famous story from a long time ago that's told in sort of human factors area is some plane. There are a lot of stories about planes and the human factors world because it's an area where a lot of study has been made about how people do the right thing, do the wrong thing and so on. But the this was a plane where there was some little tiny map light that wasn't working and these sort of three crew members, pilot and co pilot, engineer, whatever it was, were all fiddling with this map light for, I don't know, 20 minutes or something, during which time the plane descended from its cruising altitude. And I'm not sure if it actually crashed or whether it nearly crashed, but the point was they were concentrated, they were fixated on this very trivial thing that wasn't really that important to the flying of the plane. And meanwhile they managed to kind of miss the big picture of the fact that the plane was descending by tens of thousands of feet and so on. And I think that's a not uncommon thing that one sees happening with people who make when mistakes get made. It's a thing. One, it happens to all of us that you're fixated on one particular thing and you don't notice kind of the bigger picture of what's going on and then a bad thing happens. But the question is, if you're, if you're very specifically trained in one thing and it's like, don't worry about anything else, you just need to know about this. One thing I think that probably emphasizes, enhances the likelihood of that happening, I think, well, there's probably more to say about that. Let's see. Sando comments. Version numbers for products get fuzzy over time. Are you thinking about using years instead? You know, this question about how you sort of predict the future of products is a tricky one. You know, if you say we're going to sign up for calling it, you know, the 2024 edition or something, then you better not miss 2024. And in a sense there are sort of forces, I mean, like for example, it's now the end of July, we're getting out version 14.1, we're getting out two versions per year. In a perfect world, this would have come out right at the mid year point. It didn't. It's a few weeks late. A few weeks late for a big software system is really very good. But it's still, it's not sort of precisely at that time. And so I'm always nervous when one starts thinking about let's promise to do exactly this number per year and so on. I'm nervous about doing that. That's one thing. Now, the other thing people do is to name versions. We could name it tiger, Lion, Ocelot, whatever else. Somehow that's reminding me. When I first had networked computers back in the early 80s, there was sort of a question of what would these computers be called. And I had various different series of computers named after stars, named after big cats and so on. But in any case, one could do the same kind of thing for naming versions. And I think Apple does that, I think, where you just have kind of this word to describe a version. I have to say, while it worked sort of for a while for me with Apple, I have totally lost track. And the fact that there is no sequencing there is difficult. But yes, we have discussed at some length what to do about kind of the sort of versioning and what does the version number mean? And I think the main thing that's important about version numbers is they're numbers which have an ordering. And that's really the only thing you need to know. And then you need to know what's the latest version. And this version that I have is older than the latest version. So that's why we've stuck with numbers for now. Let's see. Misha is asking advice for autodidacts. How does one turn personal curiosity or questions about science into structured projects that can be published, like I seem to do? Well, I think the main thing is what's a question you want to ask? And is that question one? You know, one good test is if the question that you're asking cannot be described in less than like, 30 sentences, it's probably not going to be a generally interesting question to answer. That may be unfair, but that would be an immediate criterion of mine. Most of the questions that I've been sort of interested in asking and so on, have been maybe not one sentence, but a couple of sentences. Assuming you're talking to somebody who knows a reasonable amount about science in general, at least, probably not specifically about the particular technical aspects of what you're talking about, because one wants these questions, which are reasonably easy to state. Those are much, I think, are sort of better bet in terms of when one's able to make progress. But I think the thing is to define questions to sort of crispen up those questions, then go work on them. And for me, I mean, my particular form of learning, I tend to not learn very well unless I sort of have a question in mind. And once I have that question in mind, then I'll go and try and drill, drill, drill to find the answer, whether that's by reading books or papers, or whether that's by figuring things out for myself, or whether that's by finding people who know about that stuff to talk to and so on. And I think that tends to be my approach. I mean, some other people can sort of speculatively learn things just sort of for the sake of learning them. I tend to sort of pick up facts for the sake of it. And I'm always curious when I talk to people who are experts in different kinds of things to learn about the things that they know about. And I sort of gradually accumulate more and more facts about different kinds of things. But in terms of the things that I really concentrate on and that I really make an effort to kind of specifically study and read about, they tend to be things where I've got a question that I'm trying to answer. And that's the guiding principle, the guiding theme for the things that I look at. Let's see a question here from Jamie. Do you think AI will take away human autonomy, ultimately making humans less intelligent overall as they rely on AI too much? Well, it depends what we mean by intelligence, I suppose. I mean, I think there was a time when knowing how to find your way with a map was something that sort of seemed to be an exercise of intelligence. And some people could do it well, some people couldn't do it so well. Now we all have GPSs and it's not, you know, that piece of intelligence, so to speak, is not so widely used now. You know, if you say is the GPS tells me to do this or that, I am not using my brain to figure out, no, that's the wrong way to go, that's totally crazy. Yes, that's the case. And you kind of start relying on the GPS to tell you which way to go. Even though if you really thought about it, you might be able to come up with a better conclusion or you might be wrong and the GPS might do a better job. And I think that's one of the issues is that quite often the kind of the auto suggest to life that AI, like systems provide is sort of good enough, typically, and it's not obvious you could do better. And so you just follow the suggestion. And yes, that is a, you know, if, if there are, it's worth saying that there are things over the course of history that gradually get automated and where you don't have to put mental energy into thinking about those things yourself. But occasionally you will make a mistake because the AI will do something goofy and you were not thinking about it and you just sort of delegated it to the AI. Now I tend to think that when we provide tools, we kind of raise the level of, of the place where we humans can make a contribution. I mean, I know in my own efforts to build our computational language, Wolfram language and so on, I've kind of built this sort of taller and taller platform from which I can start operating and trying to figure things out and so on. And that has been absolutely critical to things I've been able to do with science and other kinds of areas, is that I'm using huge stack of automation and that's which just allows me to start from a much, much higher point. If I was just doing all that stuff step by step myself, I'd never get to the places where I can get to now. So I think that yes, it is true that as you allow things that you're doing to be automated, there are things where you might say every so often, gosh, I wish I hadn't left it to the automation, it did something really silly. But there are also more places probably where there's more leverage from getting to, to sort of make use of what already exists. I mean, I suppose it's like, you know, if you grow your own food, you can probably get somewhat better food in some ways. In some ways, maybe not in other ways. You don't get to grow things that don't live in the climate that you have, unless you've got a nice greenhouse or something. But you know, it takes a lot of effort to do that, even though occasionally the results might be better. And if you use kind of the whole sort of supply chain of food or whatever, it's potentially a lot easier. And I think the same is true for sort of intellectual activities. And as I say, if you learn to use the automated tools, well, then you simply get to take advantage of them. If you just let them wash over you and you just follow the auto suggests and don't make use of those things as kind of pieces of leverage, then yeah, maybe that, that it doesn't lead to as good a result. I mean, the thing that is sort of a concern, I suppose, is that insofar as people just say, oh, I'll just write that letter with chatgpt or whatever, you know, those letters aren't terribly good. And the insofar as sort of the AI is learning from sort of the mass of humanity, it's kind of reverting everything to the mean of what humanity can produce and it sort of becomes. And at the point where that becomes the expectation and where it's like, gosh, it's really weird that you're doing it this other way. Then one sort of reverted to a point where, yes, it feels like humans are not doing the things that humans could do. They're kind of just sort of going with the AI flow, so to speak. But I tend to think that if you. If I would hope that the main thing that will happen is, is sort of a leveraging of what now becomes automated so we humans can do more now. It's sort of interesting. I mean, when ChatGPT first came out, year and a half ago, a little bit more than that, it's kind of like, okay, now all these jobs just go away. We don't have to do them anymore. It'll just take over all these things. I think most people have realized that's just not really right, that it lets one do certain kinds of things. It provides a very nice linguistic user interface. It provides a better way to do various kinds of knowledge, sort of presentation and retrieval and so on. Provides a bunch of different kinds of capabilities. But it isn't the case that it's like, okay, this just goes away. It changes the nature of what has to be done by humans. It's just like before there were calculators and computers and so on. A big part of, I don't know, operating something like a bank was simply adding up columns of numbers to figure out what the accounts, what the right numbers for the accounts were. That all went away. But that doesn't mean that the banks don't employ any people anymore. They just do somewhat different things. And I think that sort of the shock of ChatGPT, which nobody expected, even the people working on it, didn't expect it to work in the kind of fluent way that it was able to work. It was just not clear in what year one would sort of get over the threshold where one could get sort of fluent writing done. And turned out to be late 2022. But nobody really knew that. And sort of the extrapolation from that is, okay, it did this, it's going to do everything. That's just not a correct extrapolation. I mean, the thing that we get to see, because we sort of provide a bunch of the kind of computation capabilities for the world, is that the sort of computation side of things is not what the ChatGPT or large language models or whatever. That's not what they do. They are doing something which is very much about linguistic user interface. Maybe Human language, maybe a little bit of code, some other kinds of things. That's what they're specialized in, so to speak. That's what they do. Well, it's different from the kind of more formal sort of stack of computational capabilities. And it's something where people sort of have the assumption, well, they can do one thing well, so they're going to be able to do everything well. But which seems, you know, I think people are realizing now that that's really not the case. Kind of good for our business, I suppose, because we happen to provide this very, this sort of the broadest sort of ability to access the world computationally. And so that's a very useful complement to kind of the linguistic user interface that LLMs provide. But I think this question of sort of do the AIs make humans stupider? I think that the answer is only if the humans choose to operate only within the things that were sort of now automated by AI. But the automation simply provides more leverage, a taller platform for things that humans could do. At least that's my view. Let's see. Brady asks, how do you think the patent system could be improved by AI? Oh, I don't know. The patent system is always, always hard for me to understand and appreciate. I mean, I think it's, it's, you know, patents are really hard to read. They're written in kind of formalized English. That really is a form of code. I mean, it really, in a sense, should be a big lump of computational content because it's like we're trying to describe this thing. And by the way, if you did insist on patents being computational documents, there will be a lot less confusion about what was actually in a patent. Maybe a lot of patents wouldn't exist because as people tried to clarify them, they would realize there's really not something very interesting to say here. But I think that's one question. Another question, that's a general question, I don't think necessarily relating just to patents, but is to what extent as things get invented, to what extent can one sort of allocate some degree of creativity, originality, whatever else, to those different stages of invention? Like, for example, when you have a thing, oh, I don't know, maybe it's something in a virtual world and people keep adding to it and people keep modifying it. And then you might try to define, is there a way that you can say this thing that's now been designed, this very elaborate machine, or something that's been designed in some virtual world that has all kinds of flamethrowers and legs and who knows what else. And people have been adding to it. And then you ask the question, well, okay, this machine somehow is getting monetized. The fact that somebody invented the foot and somebody else invented the antenna and so on, is there a way that you can reasonably allocate sort of contributions to the different folks who had made pieces of it? And I think it's a very non trivial problem to figure out given that you have sort of a developing piece of intellectual property content, whatever else. What has been difficult for the system, I think, to date is it's like you write something, boom, it's there. It's a copyrighted thing. There are concepts of derivative works, there are concepts of things like this, but it's kind of a kind of a one gulp type thing rather than, oh, there's this document and lots of people worked on it and they added little pieces here and there. You know, it tends to be the case that copyright, for example, it's like anybody who sort of contributed, there's either you contributed or you didn't. If you contributed, then you have certain rights. If you didn't contribute, you don't. It's kind of a very binary thing. And so one wonders whether in sort of a world of kind of thinking about things more computationally, it becomes possible to have some better way, some smoother way to kind of allocate responsibility for the different aspects of what was done. Now, obviously there are issues with this all over the place. I mean, there are issues when AIs take lots of training materials that people have created and they sort of eat them and then they grind them up in certain ways. How much does the ground up thing still reflect the thing that went in, et cetera, et cetera, et cetera. And people have all kinds of ornate arguments about how to apply copyright law, which was built at a time which didn't imagine AI's reading training data on the web. That really was a very different time when people were making books and printing limited numbers of books, things like this. And so the question of can you take the words that were written in the copyright law and sort of adapt them, or can you take the principles and adapt them to modern times? That's an interesting problem. And people, I think, sometimes have these little threads that they pull on, like, well, if you want to do machine learning training, you had to make a copy at some point as the thing was processed through all the different stages of the training pipeline, things like this. And does that matter based on the original sort of structure of copyright law? And so on. I think these are hard problems, but I think in terms of the general question of what might become possible, having things be much more computationally defined, instead of patents being this written in legalese, have them actually written in computational language. I think with Wolfram language we could get some distance towards writing a patent directly in Wolfram language. We're not quite all the way there yet. Actually we have a project that for sort of generalized symbolic discourse language which might end up having the result. You could do that. If you do that, what can you then do? How can you then change kind of these questions about is it a binary thing, a contribution, not a contribution, as you incrementally contribute things, et cetera. And of course then there are all sorts of questions which also are the looming questions in all these areas is what does society want about all these kinds of things? For example, patents, the original what is it? I can't quote it verbatim, but the thing in the US Constitution that sort of initiates patent law, something about giving to inventors and something in the useful arts, I think for a limited time the exclusive use of their inventions. I mean, I think it may save discoveries and inventions, but kind of there's a principle there which is that if you want people to invent stuff, you have to let them have some benefit of their effort, or that's a societal thing, or otherwise they're going to say, eh, it's not worth the trouble. Now, you know, it's been an interesting thing in science, for example, there is no analogous structure. The question of, you know, you invent something in science, people can, you know, can rip it off any way they want. And maybe that's what the world should want to be the case that a lot of things I've done have been very much sort of open science. And it would be, in a sense, at some level one needs an ecosystem where the doing of that successfully can pay for more of the doing of it, because otherwise you just won't be able to do it. And there has to be some sort of financial piece to the whole story or it's really hard to keep doing those things because in the end the people who do it have to eat and so on. But I think this question of the societal aspects of what does one want? That's one question. Then there's the question of can you implement something where you can say, well, many people can contribute to this and they'll all get their fair share of what comes out. We can't really do that right now. And that would be a thing. In people's discussion of blockchain and so on, that's something where one might imagine that was a thing that could be something that that kind of thinking of tokenization and so on could lead to. But I don't think that's really quite gotten there. And this question of when you have a thing that is described by a piece of. So, for example, let's say you have a patent, and it's really described as a piece of computational language. And then somebody says, well, there's this extra piece, or there's this piece that was already figured out. And then you say, well, is that the heart of the idea? And you say, here's another piece of computational language. That's the thing that was already figured out. Now, as you look at the big patent, does it, you know, this piece that already existed, let's say, you know, is that really. Does that eviscerate the whole patent, or is it just, oh, that was a piece, and then there was more added to it? I think in patent law, there's this thing that usually gets called otsog, the principle that I think it stands for on the shoulders of giants. And it's kind of a quote from Newton where he said, if I have seen far, it's because I have stood on the shoulders of giants, which many people think was kind of a barbarian at Robert Hooke, who was a person of very short stature and sort of a competitor of Newton's in some way. But in a sense, this principle has the idea that nothing is ever completely new, ever starts from nothing. Things start from things that already existed, and then they build on there. And in a sense, if everything was completely new, we wouldn't even have words to describe it. We wouldn't be able to. Something that is completely new in some sense, is so new that only maybe the person who originates it knows about it, and it's kind of not even communicable. But I think then the question is, okay, so let's say you had a computational version of patent A and patent B. And you're asking, how important is patent A and patent B? Well, you might have a fighting chance of figuring that out with the computational version of these things by some criterion. The criterion could be run simulations of what you can produce with these things. What overlap do you have between this and that? I'm not sure something would require thought, but I think it's something where there is a potential for some sort of new direction beyond what exists today. Let's see. Corky is asking I wonder if we go through a cycle of trusting AI Far too much for answers to our questions, and then get to the point where we have incorrect information, we don't trust it at all, and then get to. And where. Where will we end up? Yeah, well, I mean, look, it's like. It's like kind of hype waves about lots of kinds of things. You know, at some point, everybody's got to do this. Some, you know, new health, fashion, or some new kind of way of leading one's life, fashion. And people get very, very extreme in doing that. And then people realize, wait a minute, that's too extreme. It has lots of bad effects. Let's back off from it. And it kind of porpoises around until eventually it reaches some equilibrium level. And I think that's probably the same thing that will happen with kind of trusting AI as has happened with lots of other kinds of things. I think it's kind of a human tendency that people say, this is cool new thing. Let's all think it's the greatest thing ever and concentrate on that. And then there's kind of an overshoot, and then it goes back and so on. My guess is. It's going to work that way. Let's see. Kaleida asks, what has been your favorite place or country to visit? Is there some place you have yet to visit that you would like to. Wow. Well, I've been very episodic in my traveling through my life. There have been periods of a decade where I've traveled, really not at all. And other periods where I've traveled a lot. I think one period when I traveled a bunch was about a decade ago, when particularly my older daughter was a teenager, youngish teenager, and was interested in traveling and knew that I got all these invitations to all these places all over the world and. And I turned them all down because I wasn't traveling. And she said, let me just read through these and try and, you know, and decide which ones we will go to, so to speak. So. So that got started and traveled to all kinds of places. At first, sort of driven by invitations. I think we wound up on an island off the coast of Australia within a few weeks of her. Let me read through the invitations and see which ones we're going to go to, so to speak. And in any case, the thing that after a while started realizing, well, there are places that seem like they'd be interesting to go, and it turns out there are lots of places where there's. In any given place. I'm not a very good Tourist as such, my main interest is usually the sort of human and organizational side of places I'll visit rather than the pure tourist attractions. And so it's usually I find, oh, I don't know. Universities are interesting, but not as interesting as they might be because, like, research universities are fairly similar all over the world. Companies, depending on the kind. They're sometimes like, we're trying to be just like American startups, or they can be rather different. Government can be somewhat different, schools, like high schools and so on. As you go lower and lower in sort of the education stack, I think things get more and more different between different countries. And that's pretty interesting. And it's always pretty interesting to talk to folks who are not in the sort of fully globalized set, so to speak. Then you actually get to learn something about different ways of thinking about things and so on, at least for me. And there occasionally are things where it's just like, I want to see that. That strange thing that exists in some country and not others. I suppose that's tourism. But my forms of tourism are pretty weird and sort of revolve around sort of scientific and science and technology kinds of things. Much more so than the typical kind of interesting building or whatever else, you know, I do. Perhaps I can tell a story about my fine family, which not all members of which were. Did the kind of travel to lots of different places. And, you know, I tended to sort of get an intense schedule of visits to various kinds of things. And at least two of my kids, I think, found those pretty interesting. But my wife and my other two kids, but we're not really into those things. And at some point it was kind of like, was an idea, where can we go on a sort of family vacation where there'll be no way that I'll be able to have any kind of sort of meetings and so on. That can be done. And so the place that was picked for that was Svalbard, which is northernmost kind of populated place. I think it's at 78 north latitude, maybe. I think it's the last place on the Earth that's still governed by the Treaty of Versailles. It's a cold version of what I sort of imagine the Wild west might have been like. Although I think maybe it's been transformed since then because the harbor is deep enough that cruise ships can go there and stuff like this. But in any case, it was sort of amusing that I discovered when I was there that there was a group that worked on ice physics that happened to be users of our technology. And so even at that level of the ends of the earth, it turned out there were, well, it was very, well, interesting visit for all kinds of reasons, but that was an unexpected piece to it. But in terms of places, I'm always curious, but it's often one sort of just doesn't get around to going to places. And it takes effort to organize the kind of visits that I found interesting. I have often sort of. I suppose they're places which I've considered visiting, but they're really too much trouble. I mean, they're places I'd be really curious about. You know, go meet folks who are far away from, you know, indigenous tribes in the Amazon jungle or something. Far away from typical. The typical experiences that I have. But it's kind of like so much trouble to arrange for that. And it's not clear what would happen. And it's kind of like, you can't be serious, you're going to go there. That really doesn't seem very safe type thing, et cetera, et cetera, et cetera. So I've been, I've been. I've not done those things, but I think that's, I mean, it's, it's one of those things where it's a question, what does one get around to? And also, you know, I tend to be a, you know, I work hard in a lot of projects and kind of just going on some trip somewhere while it's interesting, it's kind of always quite disruptive to the things that I'm doing. And so I kind of have to trade off as it's so interesting that it's really worth doing, even though, you know, it's, even though it's disrupting these other things I'm doing. I think one criterion is, you know, I've been to lots of places and most of them I could say, you know, yeah, I remember that. That's interesting. I learned a bunch of stuff from it. And some of them are like, oh, did I go there? I can't even remember if I went there. And those are the ones I. Those are the trips I shouldn't have gone on. Let's see, There's a question here from sk. What is it like starting one's first business? I'm wondering because I don't personally know anyone who has a business that's interesting. I mean, in different parts of the world there will be different kind of profiles of who's starting a business. I mean, if you go to Silicon Valley, for example, and you go to a Starbucks in Silicon Valley, and You listen to conversation fragments. You'll hear, you know, MVPs, Series A, you know, all these kinds of these phrases that come from the startup world are, are sort of thick in, in the conversations that you'll hear the fragments of, I think, you know, so it's kind of like, and I find it amusing in a sense that in sort of Silicon Valley at this point, you know, the, the high school students are like, yes, I have a startup, et cetera, et cetera, et cetera. That's not true in most of the rest of the world. I think the, this question of, of what's it like to start a business? I mean, I would say the main thing to realize is that it's sort of all on you. That is, if you're in some setup where you're in school or you're in some existing organization, it's like somebody's going to tell you what to do. If you are starting a business from scratch, it's just you, Nobody's going to tell you what to do. And that for some of us, like myself, that's great. For some people, that's very disorienting. It's like, well, you mean, if I don't get up and go to work this morning, nobody's going to, you know, nothing's going to happen? Well, maybe my business won't succeed, but it's kind of like nobody's telling me to get up and go to work type thing. It's something which is kind of, it's all on you. And I think that's a feature of starting businesses that people, for some people, that's something that's like, great, I just get to be in charge of my own destiny. And for some people, it's like, oh my gosh, what do I do next? Nobody's telling me what to do. And in a sense, there's been some attempt to kind of scaffold that by saying, you can go to a class, tells you how to start a business, and perhaps that for some people will help. But in the end, it's kind of, you're there, you're doing your thing, nobody's telling you what to do, you're deciding what to do. It's all on you. So I think that's a key aspect of it. And I think there's a tendency for people to say, let me go ask other people what I should do. But the thing to realize is, if you're starting your business, it's your business, and the way you do it will probably be different from the way the 10 other people who might be suggesting how you should do it will tell you how to do it. You know, you really have to, it's a very, it's a, it's kind of a, a sort of a personal self discovery type thing, starting a business because your business will reflect you and if you try and make it not reflect you, it probably won't work out very well. And so you learn a bunch of stuff about yourself, good and bad, by starting a business, so to speak. And I think it's important, I think it's important to be sort of true to yourself in what you do with the business. And sometimes people will be like, well I should be like this, but I'm really not. But I try to make my business like the thing I should be like and that often doesn't work very well. So again, it's one of these things where it's going to be sort of an extension of you and it's going to sort of amplify aspects. You know, if you're really good at this kind of thing, your business will probably be good at that too. If you're really bad at this kind of thing, it's pretty hard to make your business not be bad at those kinds of things. And so I think those are, you know, those are, those are aspects to what it sort of feels like in a sense that it's, it's kind of a, well it's, it can be a little bit, it really is a reflection back on you. Let's see, Elsie is asking, if you had to start an innovation intensive business nowadays requiring R and D before revenues, would you go the VC route or find ways to bootstrap? Well, I think the VC route is the last choice. I think you may have no choice, but I tend to think it's the last choice. Even though I have many friends who are venture capitalists and I have many friends who are both very serious, well known venture capitalists. And saying I'm an investor, I'm a venture capitalist seems to be one of those generic things that people say if they're kind of, you know, a little bit retired from other things they're doing and so on. So there's, there's two categories there. But I think the issue is again, it's a question of, you know, who wants to get what out of what's going on? Who has the vision, you know, who. To what extent do people want sort of self determination? To what extent do they want somebody else telling them what to do? I think as soon as you have investors, you are responsible for the money they invested, and you know, you should do the things that, that sort of, in some sense they want you to do with their money. It's a different thing. If, if you don't have their money in your business, then you're sort of free to do what you think you should do. So I think, you know, I, I tend to think as a practical matter, so, you know, I, I've not been on that side of, of dealing with these things for a very, very long time. But certainly the complaints that I hear repeatedly from people who sort of get venture capital, you know, there are not that many people who, after a long period of time, at least, well, they tend to have trouble. And you know, they have trouble because either some big venture capital firm has some young hotshot who's absolutely sure they know how to run this business better than the person who is, whose vision that business is, but they don't know what they're talking about, or they're something where basically the interest of the venture capitalists, which is just outright making money, are not well aligned with the interests of the people who are running the business who might be interested in kind of what the business makes. I mean, it's kind of like one is about the money the business makes and the other is about the stuff the business makes, or usually is in the most, I would say most successful businesses, the people who at least found them care about the things those businesses are doing and aren't just doing that for the sake of making money. It's perhaps a little bit different in the finance area as such, but that motivation there, I think is more in the problem solving. And it's kind of cool because we make tons of money type thing. But I think in it's this question of how do you bootstrap things and so on. I mean, it's challenging. I think the best thing and what I did with our company 37 years ago or something now is find customers who want the thing that you're trying to build and who sort of are excited enough about it that they'll pay for it or at least part of it in advance. And you know, that way it's like you're making something, you're signing up to make something which has certain value and people are paying for it, and they trust you enough that they're prepared to pay for it in advance because they just want the thing. And I think that's a good way to do things. I mean, there are lots of programs of whether in the US the sbirs for small businesses and so on. My observation of those is that they tend to be rather distractive because they tend to have therefore something very specific and the thing you're doing probably isn't that very specific thing. And there are other sort of channels of depending on what kinds of things one's doing. There are things that are more sort of nowadays, sort of mostly tech industry people who have turned to various kinds of philanthropy and who recognize that starting companies is a good way to make certain kinds of things happen in the world and are prepared to sort of invest by giving money rather than by sort of becoming, taking, you know, investment money and so on. So I think those are, those are a few mechanisms. But it's, you know, it's also a question of to what extent one is concerned that the thing one is doing is a thing that only one understands and to what extent it's kind of a thing where, yeah, people know what this is. You can kind of make these deals. Maybe somebody can be an investor and it's part of some portfolio that they understand. Maybe somebody is sort of made some, you know, if you're out on a limb with something very innovative that only you believe in and so on, it's much more challenging to kind of get, get all that stuff set up. The. So I mean, I would say that, that, yeah, you, you will almost certainly lead a happier life if you, if you don't have sort of other people telling you what to do at least. Well, that's based on at least my personality. I mean, some people are much more comfortable when somebody is telling them do this, then do that and so on. They find it disorienting and, and kind of, and, and sort of scary when it's like it's all up to them, I think. Yeah, let's see. Chow asks, can AI systems be effectively applied to customer support roles or is there too large of a security vulnerability? Yeah, it's an interesting question. I've interacted with people about that issue. You have an AI system. It's doing customer service. If the customer service capability includes send me a replacement product, send me a new car, send me a new phone. If you can kind of convince the AI to send you a new car or whatever it is, and people on the Internet discover that there's this way of having this whole conversation with the AI and then you can, pretty soon a new car will drive up. That could be big time problem for a customer service AI. I don't think it's so difficult to avoid that. I Think the main thing you have to do, which is sort of a general security practice, is the thing that's at the front lines. Talking to the customer is not the thing that will ultimately send the car out, so to speak. But you have another thing that the customer can never talk to that is for example, another AI that is watching what the first AI is doing and is saying is something crazy happening? That's I think at least a way to remediate the issue of yes, there'll be a hack found that convinces the AI to do something crazy and so all is lost, so to speak. Like most kinds of security issues, it probably won't be perfect. And I think that's sort of the price you pay for having something with the sort of computational power of AI is occasionally it may surprise you, perhaps negatively. And do you need an escape valve if all goes wrong? Well, a lot of human systems have escape valves. If all goes wrong, it's kind of like refer it to the manager or the supervisor or the whatever or the supervisor of the supervisor or whatever. And it's sort of the same kind of thing in this case. And I just the exact way that works. How will sort of all the protocols get built and then what level of oh my gosh, it was a crazy thing that the AI did. Is one prepared to accept remembering that the fundamental point about a lot of what's going on with these sort of computationally sophisticated systems is if you want them to be computationally sophisticated, you're not going to understand the details of how they work. So saying we ought to make sure guarantee they'll never do the wrong thing. It's not realistic to do that. If you want those systems to sort of be able to do computationally sophisticated things, that's kind of the trade off and I think the implicit trade off the world will end up making is to let's just delegate a bunch of stuff to the AIs and let's try and put in place systems to sort of remediate any issues. Now the same is true when you have networks connecting computers. Those networks in principle allow people to go and sort of worm into your computer and do bad things. But people just decide having their computer on a network far outweighs in value the possibility of somebody coming and kind of attacking their computer. Except in certain rare cases where the sort of cost of an attack is so high that you have to have the thing perfectly air gapped and all that kind of thing. I think that that's. And so my guess is that's sort of the way things will go, and this question of, okay, well, we've connected all these things, now we have to have this whole infrastructure and ecosystem of computer security that tries to sort of patch things up. So, yes, we are connected, but we're not connected too much. And you can't get into this, and you have to check that and so on. And that's what I would expect will build up around AI as well. Um, let's see. JD is asking, can you elaborate on your experience expanding your business and products to be used by others whose languages you don't speak? Well, gosh, there's several pieces to that. I mean, one is, can you sell stuff in places where you don't understand the language or more importantly, the culture? That's challenging. It's. It's. You know, usually, often one goes in two steps. You find somebody who will resell or work for you in place or culture X, and you have to interface with them, and then they interface with the local folk with that particular culture. But that can be challenging because even interfacing with the interface person, they will be at least halfway, if not all the way into that different culture. And you really, you know, the expectations are very different. I mean, in some, you know, in. In the US There's a certain style of doing business where people are fairly direct in certain ways, not so direct in other ways. People expect that there'll be legal documents and people will sort of follow the legal documents, all this kind of thing. Whereas in other parts of the world, people may be unbelievably indirect. And, you know, you'll go through a long business meeting and everybody is just doing something ceremonial. And the only thing that really matters is the two sentences said in the last minute of the meeting. For example, that's not typically the way that things work in the US and it could also be the case that you think that it's something where instead of it being, yes, everybody really concentrates on this contract, and then we just follow what the contract says. It's more like, well, we all understood this and that and the other. And it's more of a kind of a personal agreement type thing or something where you're selling something and it's all about kind of personal relationships. And it's not just, well, this is a thing, and we buy this thing because we find it useful for this and that. It's about, oh, yes, I know this person, and they're the kind of person that I think I should be buying something from, and so on. I mean, just to Give a sense of some sort of cultural issues and translation issues. I remember when Mathematica was young, back in the late 80s, early 90s, we were translating the book I'd written, the Mathematica book. It's a documentation for the system. We were translating it into Japanese and we had a translation and there was a user group, there still is a user group in Japan for our products. And the user group was like, your translation is just wrong. It's just the wrong kind of translation. And it's like, well, is it Japanese? Is it correctly Grammatic? They said, yes, it's grammatically correct, but it's just the wrong translation. Okay, what did they mean? Turned out, as I sort of learned, there were two ways that you could have something be presented in Japanese. One was a very authoritarian way, the way that textbooks were often written. The student should know, blah, blah, blah, blah, blah. The other way was the way that commercial documentation was often written, which was, if it may please the customer, do X, Y and Z. And these two things, apparently in Japanese, I don't know from firsthand experience what I've been told. These two things really lead to very different forms of document. And so we had done. I forget even which one we'd done, but we had done one and everybody said it should be the other. And eventually we got something which was sort of a compromise, but that's kind of a gotcha of doing, of even communicating what you have in a culture that isn't your own. But I think the challenge is usually to have these kind of steps of do you go from the thing you understand to the thing you're trying to sell to, so to speak. I think that the specific issue of. Of kind of language translation, really, that's the thing for which transformers, which led to large language models and so on, that's what they were invented for. And they do an increasingly good job. I mean, it varies with how similar the languages are and how big the training corpus is. Because if you're in the EU or something, every document in the EU is translated into all the EU languages and is a public document. So there's a lot of kind of parallel training data, even though it might sound rather legalistic. Or for example, computer manuals, there's sort of a group that collects sort of computer manuals that have been translated into many languages to use as training corpuses for AI and so on. So I think at this point the pure language barrier is definitely coming down. I mean, I have yet to do a zoom call or something where I'm talking English and somebody else is talking a completely different language and it's being automatically translated. I don't think that capability is quite here, but it will be soon. It is the case. I know somebody ran an interview with me, I think in Portuguese, where it sounded like me, but it was talking Portuguese, and I was told it was a pretty decent rendering. And that's sort of the beginning of really kind of smooth, just purely language translation. But the culture translation is much more difficult. And in terms of selling things in different places around the world, it's, you know, I think I consider it somewhat challenging. And I mean, it's certainly the case in the sort of geopolitics of the current world that a lot of the large markets are really making it very difficult, for example, an American company to sell there, and possibly the US Is making it. Might make it. I think the US has been doing less of that, might make it more difficult for those countries to sell in the U.S. but it's, you know, some of that that is becoming increasingly difficult. I would say. I mean, I would say that in the distant past, you know, in. In, let's say, up to the 1960s or something, 1970s, even the international banking system wasn't really well established. And, you know, when you were trying to work with some company in some other place, you know, people were sending physical letters back and forth, and there were all these letters of credit, and there were all these ways that you would get this kind of chain of trusted, you know, people or entities along the way to get to that other company that was in, you know, somewhere in South America or something, and you would go through a chain of things. So that sort of everybody trusted everybody along the way. And then things became, particularly with sort of globalization, you know, international banking and things that all became a lot simpler. I would say that there's a reversal of that in recent times as, you know, a bunch of countries are like, wait a minute, we don't, you know, we'd like more stuff to be developed within our own country. Makes some sense. You know, one can. It's. And I think so things are becoming more difficult again. The last probably five years, things have become significantly more difficult again after a period when they were comparatively easy. Let's see. Misha is asking, I'm curious what your favorite revelation or idea you've read in your recent dive into philosophy. Oh, I don't know if you should really ask that. I. I've been. Yes, I've been trying to learn about a lot of different areas of philosophy. I I suppose I. I'm a little nervous to tell you one that I thought of just recently because it might not be right, but I was trying to learn a little bit more about Buddhism and kind of the view of sort of reality from different points of view. And it's kind of sort of this question of what we ultimately perceive is kind of somehow deep inside our mind. And there's the outside world that has things going on in it, and there's a sort of layer in between of what kind of we. Of how we process sort of the outside world to bring it into our internal perception. And there's sort of the question of how much machinery is there in that processing, how much is it inner things within our minds versus how much is it what's going on on the outside? And I think some of what I've studied with the Ruliad and the way that we kind of exist as entities within the Ruliad and so on has to do with this question about the trade off between what we are sampling versus what's out there and also to some extent what we are constructing beyond just what we are sampling. In other words, to what extent do we have an inner life of things going on that might be only loosely connected to what's in the outside world? And I suppose one of my. You ask, and I'm telling you, I'm not necessarily standing by this, but I was trying to understand the kind of concept of enlightenment, so to speak, since you've asked about revelations and things, the concept of enlightenment and kind of Buddhist philosophy and the question of whether what's going on there is that that's a point at which you have enough inner stuff happening that it's kind of like, forget the outside world. It's just like the inner thing is kind of all I need. And I suppose that whole question of whether the inner thing is all you need, it also relates to, at the point where you have a sort of a more sophisticated mind than our human brains. To what extent can you be sort of running the simulation of the world right inside your mind and so on? Maybe this is not the right topic for, for today. Maybe just one more practical question. I'm sorry, I'm not necessarily going to stand by that. That's my last three days, sort of proto understanding. So it's very raw. I mean, I have been thinking quite a bit actually about. Well, as I've been working on some practical issues about kind of understanding how machine learning really works and what really allows it to work and understanding what. What I'M discovering means for how brains might really work and so on. And sort of I've been thinking about questions about sort of the transferability of thoughts from brain to brain and how that works and what kinds of, sort of more sophisticated levels of thinking could exist in kind of, sort of beyond brain's version of mind, so to speak. And I've been making some progress on this, which hopefully I'll be able to talk about soon. The. Let's see. All right, one more thing, maybe. Okay. Holly is asking, how often do you revisit your own personal goals in life and in your career? What are some things you look at that make you feel accomplished, whether small or big? You know, I'm about to have a. A disgustingly large birthday. And so I was. I was thinking what I might do a live stream and talk about some of those kinds of things. But I was thinking maybe I should be looking back. I've had a particularly interesting last five years. When my last big birthday, so to speak, five years ago, I sort of said, hey, I'm going to try and actually finally get around to doing some things I've been meaning to do for decades, like figuring out things about the fundamental theory of physics. And to my great surprise, the efforts that I put in that direction have really worked out really well. And in fact, I've been quite on a roll kind of doing things, foundational things that I'd sort of meant to do decades ago, and I'm finally getting them done, and that's kind of satisfying. And so I was sort of looking back on the last five years of a time when I managed to sort of close off a bunch of things that have been sort of open for four decades or more. That's a thing where, I don't know. The good news is I'm not running out of things to do because as every door that closes, so to speak, it opens another one. It's something where I think the thing that there are sort of new things that become possible on the basis of things that have been done. But in terms of. Overall life goals and so on, I suppose it's like, am I doing very much the wrong thing? What should I be doing, so to speak, at this point, I don't think I'm doing that well. I mean, I always try to think, am I doing the wrong thing? And am I sort of putting emphasis in a place I shouldn't be? And I suppose I feel like the right way to lead one's life is to spend some number of percentage points of time Planning what one's going to do, and the vast majority just doing things. But one shouldn't zero out the planning time. It's always worth spending some time saying, what's the big picture of what I'm trying to do? I suppose that's another thing important to me is one can get very detailed in some rabbit hole of minutiae of what one's doing, but the question is to sort of at what point one zooms out and says, hey, what am I really trying to achieve? What's the big picture of what I'm trying to do? And I think that's a pretty useful thing when one's doing things with companies, with research and so on. Is the zooming out, what's the big picture? And perhaps is this a thing I shouldn't be doing? Should I be doing something different in terms of what makes one feel accomplished? I think everybody is very different in that regard. I remember when I was a kid, I thought. The thing that I. I started publishing science papers when I was, what, 14 or 15 years old. And I always thought it was really neat that you could like, write something and then, you know, you wait a while and you go to any university library anywhere in the world, more or less, you go to some dusty shelf, you pull it down, you could find the thing you wrote. I thought that was really cool. I don't think that's a, you know, that's one of those things which, you know, that was. Personally, for me, that was something that I thought was really cool. These days, of course, you put your thing on the web and anybody can find it, so it's a bit of a different story. But I suppose for me, well, I like getting projects that are somehow finished and where I can sort of say, this is the whole thing. I've, you know, written this thing, I've created this thing. It's like you put it out there, it's sort of tied in the bow, in a bow and done. Now, of course, I've been running a company, one particular company for, well, cluster of companies for what, 37 years now. And that's never finished. So in a sense that violates that principle. But it has many sub pieces, like the version 14.1 that we're about to release, which is a sub piece that becomes a finished thing. I suppose for me, I like understanding things. And so for me it's an accomplishment when I finally get to understand something I've been trying to understand for a long time. I think that it's. I would say I personally am Less motivated by some of the kind of typical worldly signs of accomplishment that that one has. But that's, you know, I fortunately, have been decently successful in those regards, so maybe I would feel differently if that wasn't the case. But that hasn't been. It's kind of like sort of enough of this and that, and it's kind of like now. And, I mean, I think one of the things that's also a challenge is at what point do you say it's enough? I've, you know, I've got what I want in that direction. I don't need to go any further. For me, because I'm really interested in intellectual kinds of things, particularly. It's sort of a bottomless pit. There's always more stuff I want to figure out. There's always more stuff I want to do. I think it would have been a disappointing life if I'd gotten to some point and said, this is my one objective. I've achieved that objective. Now what do I do for the rest of my life? I mean, I was lucky enough. I had wanted, from when I was about 10 years old, to be a physicist, whatever that means. And by the time I was 20 years old, I'd managed to sort of accelerate things to the point where I sort of was a physicist. And I got, you know, that's how I made my living was being a physicist, so to speak. And I suppose at that point I kind of realized, you know what, I could do this same thing for the rest of my life. And I, you know, I was, I guess I was still at that point, young and. And kind of, you know, very sort of poking into different kinds of things. And so I kind of started looking at sort of what are the other, you know, what. What doors have now been open now that I've achieved this particular thing. It took another, you know, 30 years after that before I sort of return more than 30 years, 40 years before I sort of returned to being more seriously a physicist at some level. But I think that's a. It's kind of a funny thing because you say, I want to accomplish this thing. Then you get it, you've done it. It's like, okay, that's cool. And it's kind of like a bit boring if that's what you kind of coast on for forevermore, so to speak. And I think it's, you know, I've certainly seen people where the arc of their career has been. They did this really cool thing when they were 27 years old, and then it's Kind of all downhill from there. And it's kind of a, it's always a good thing. I think it's been a good thing in my life. Perhaps it's a good thing in my life that he says rather sort of peevishly that people haven't understood more about the things that I think I've managed to do. Because in a sense, if it was kind of like, okay, there are all these cool things I did in the 80s, some of them have been pretty well understood, but many of them have not been well understood still. And, you know, I'm pretty confident that in 50 years, 100 years, whatever, depending on sort of how the world develops. But the, I think the sort of inexorable direction that, yeah, people will understand, those were really cool. But in a sense, if it was kind of, if that had been front loaded and it's like, okay, that's all really cool, then it would be somehow that would not actually be a satisfactory a situation. Even though it is a bit frustrating that one's sort of seeing these things where it's like, gosh, Will, don't you understand this? And, you know, I know it's going to work this way. Just, you know, don't, don't ask me again, you know, is it really right, blah, blah, blah. Yes, it's, you know, that's the way it's going to work. But people don't understand that broadly yet and it'll take a while. But I think the, you know, this thing where you do your great thing, everybody understands it and then you're done and what do you do for the rest of your life? And you kind of have to hope you can hit another jackpot. And that's very, very hard to do. And it's something where I think one should be not looking necessarily for the one and done accomplishment. And then what's one going to do next, so to speak. I think it's very, at least in my life, although it's frustrating in some ways, it's sort of nice to have this whole series of things one would like to be able to do and to be in a situation where it isn't like one's getting feedback that's like, hey, that's really cool, we're done now. Rather, there's still sort of a hill to climb, so to speak. And talking of which, I need to go back to my day job and do things to prep for the release of version 14.1 of orphan language, which hopefully many of you will actually have a chance to sample. So thanks for joining me here and see you another time. Bye for now.