WEBVTT STYLE ::cue(v[voice="FCD7C008-3166-40FB-9161-C613F6473423"]) { color: #DB34F2; } ::cue(v[voice="25208755-38E1-4DFD-858C-B9ACCA89EDB7"]) { color: #DB34F2; } ::cue(v[voice="5AD0886A-5EB2-483E-98DF-414176949748"]) { color: #00D2E0; } 1 00:00:00.000 --> 00:00:20.480 [Music] 2 00:00:10.500 --> 00:00:15.660 This is GamesItWork.biz, your weekly podcast about gaming, technology, and play. 3 00:00:15.660 --> 00:00:19.160 Your hosts are Michael Martine, Andy Piper, and Michael Rowe. 4 00:00:19.160 --> 00:00:22.759 The thoughts and opinions on this podcast are those of the hosts and guests alone, 5 00:00:22.759 --> 00:00:28.280 and are not the opinions of any organization which they have been, are, or may be, affiliated with. 6 00:00:26.500 --> 00:00:29.079 [upbeat music] 7 00:00:33.120 --> 00:00:35.560 This is episode 543. 8 00:00:35.560 --> 00:00:37.560 Rent a anything. 9 00:00:38.500 --> 00:00:41.079 [upbeat music] 10 00:00:47.500 --> 00:00:54.079 Hello, welcome again to games that work. Is your weekly technology podcast? My name is Andy Piper 11 00:00:54.320 --> 00:01:11.519 We are recording at a slightly different time this week than you should now Nobody who's listening to this podcast would know that because we don't broadcast it live and you get your podcast from your regular Podcatcher any time and listen to it anytime so 12 00:01:12.200 --> 00:01:17.280 This is just some scenes setting for you. I am joined during thereafter noon 13 00:01:17.540 --> 00:01:25.939 No morning during their morning on my afternoon instead of their lunch time in my evening by two michaels Mr. Michael were I and mr. Michael my team 14 00:01:26.659 --> 00:01:31.239 Michael wrote how are you? Hey, Andy. I I'm just doing picha keen and fine 15 00:01:31.900 --> 00:01:40.420 And I I you know one of the things is you know the peak behind the curtain is so important to build your audience in this world of the creator 16 00:01:40.620 --> 00:01:44.400 Economy and so Michael Martine. How's your audience? I? 17 00:01:46.180 --> 00:01:47.480 I can't tell. 18 00:01:47.500 --> 00:02:05.060 I think it's due to or YouTube or could be anything, but as Andy said, we've got a really intriguing show here for us today and a tree show a intriguing show what we'll see could be could be both. 19 00:02:05.060 --> 00:02:17.579 We're going to start off with an article, an article, excuse me, it references a Harvard business review article from Alan Pringle who writes that engine. 20 00:02:17.500 --> 00:02:44.500 We've seen this, probably in our personal experiences here recently, and I'm curious to know how much of this you guys are doing yourselves and what your thoughts are. 21 00:02:44.500 --> 00:03:17.460 interesting. And in the day job, I was, uh, we're working with somebody recently on a business problem, and we decided to create a create an agent to help us pull some data together and ended up using a code generating CLI term tool. And literally in 10 minutes, we wrote an application that did everything it needed to do. Then I looked at the code, and 22 00:03:17.500 --> 00:03:24.020 was surprisingly well-documented, et cetera. But I then used another agent against it and said, 23 00:03:24.020 --> 00:03:46.819 look at security. Anyway, oh, no, no, no, this is, this is really, really bad. And I, I saw this, this, uh, toot from, from Allen. And, and my reaction was, yeah, if you don't understand the development concepts, and you don't have a history in that, you know, vibe coding your way through stuff, you won't catch these types of problems. Uh, and so this 24 00:03:47.500 --> 00:04:13.759 did very well. However, as we'll get to later in the show, um, how long will this be the case, right? You know, it's, it's, it's interesting the additional work that's being generated by vibe coding. But how long will that be there to drive additional work? Or are we going to see some future coming up where even that is resolved? And then we have another problem. 25 00:04:13.759 --> 00:04:17.139 Are you saying that we need to have a telephone sanitizer? 26 00:04:17.500 --> 00:04:47.459 We do today. We absolutely do today. Today we absolutely need that because the code is correct. He is an agent to go do that. I did because I understood development and development hygiene and security practices and architectures and the implications of the code that was being generated. If you don't understand that and you just vibe it, you're introducing the opportunity for lots of challenges and problems. 27 00:04:47.500 --> 00:04:52.500 What if you had like, you know, several agents to do a code review, so you've got one agent. 28 00:04:52.500 --> 00:04:55.500 Oh, this one, the one I was using does do code review. 29 00:04:55.500 --> 00:05:00.500 Ah, I see. Okay. So it wasn't just security. It did other pieces as well. 30 00:05:00.500 --> 00:05:04.500 No, I specifically asked for security after it had done its code review. 31 00:05:04.500 --> 00:05:13.500 Oh, right. Because I understood as it as somebody with a history of architecture and development and security and all this other stuff. 32 00:05:13.500 --> 00:05:17.480 I understood, just do the defaults, even if 33 00:05:17.500 --> 00:05:43.579 it's a code review isn't necessarily going to address all the potential issues. 34 00:05:43.579 --> 00:05:47.500 on kimito.net and a... 35 00:05:47.500 --> 00:06:08.540 it itself looks a bit AI generated in the sense that it's got an clearly AI generated image at the top which lots of sites have now, but it talks about how we've seemed to have gone down this path of trying to replace engineers every decade or two. Certainly, when I started in industry 36 00:06:09.660 --> 00:06:17.480 in the late 1990s out of university, there was a big push towards what we're called then fourth-generation languages. 37 00:06:17.899 --> 00:06:30.500 And we've done some of them, right? So, so the idea there was you would draw your object model out in some kind of graphical tool and then the tool would generate the business logic and code for you once you've sort of given it some rules. 38 00:06:30.899 --> 00:06:39.500 And it would write code in cobalt and see because it was primarily targeting IBM systems actually as it wasn't put about. So you must have worked on visual small talk. 39 00:06:39.899 --> 00:06:46.500 Visual age small talk. So I came onto small talk, but it was pretty small talk, actually a pretty visual age actually just. 40 00:06:48.100 --> 00:06:55.100 So yeah, it was this article talks about the 1980s with case tools, which is exactly this sort of thing. 41 00:06:56.300 --> 00:07:07.699 It was it was a case tool that I was using at the time, and I got very frustrated as a person interested in tech that it was doing all this code generation and I couldn't actually get at the units and make tweaks to it myself. 42 00:07:08.199 --> 00:07:17.500 Anyway, I found the story this this this article quite interesting to read through as well and think about how we keep seeming to try to automate away things that are innately. 43 00:07:18.500 --> 00:07:47.199 FOMO complex and with FOMO edge cases and you might expect. Well, that that's exactly the point the Harvard business review makes by saying over time this rhythm raised expectations for speed and then it continues that many workers noted they're doing more and feeling more pressure than before they use AI, even though they're supposed to save time and have pressure reduced up so this have an interesting dichotomy between the two. 44 00:07:47.500 --> 00:08:17.399 And from a case tool perspective that's one of the places where I got my career start to I work for seer technologies, which developed a case tool that when you use it it would generate the code that could operationalize you know, our demo environment was a car rental, you know experience and we would use entity relationship diagrams and other materials and feed it into the case tool and that would pop the code. There you are. And thought about that for a long time. 45 00:08:17.500 --> 00:08:47.500 There we go, but there's a different perspective on all of this right. Yeah, and and I actually saw a friend of mine who is a former distinguished engineer, former security developer just incredible technical guy ping me this article from Matt Schumer on. Kind of the acceleration that's going on with case with AI tools. 46 00:08:47.500 --> 00:09:17.500 I'm going to make a statement here at the beginning which is going to highlight a prejudice which I now have, which is that this article has a following me on X link at the top and at the bottom. Yeah, I won't do that. And well, it's not that I won't do that. It just already kind of categorizes this article and the writer in a way that is not framing them positively for me as a reader. 47 00:09:17.500 --> 00:09:47.500 This article is hey, he's running a new AI startup right. So of course it's going to be AI positive centric article. But the interesting part here was looking at the growth and kind of the sea change that happened at the beginning of this month. When it comes to AI tools and models out there. 48 00:09:47.500 --> 00:10:17.500 And he goes through kind of 2023 2024 2525 and when we started and I remember doing this, I was working on an internal project at my old company and comparing our internal models, which were hyped dramatically and did a horrible job generating code and chat GPT at the time, which would do stuff. But it wasn't great. 49 00:10:17.500 --> 00:10:47.500 We just talked about on the last article were there in spades, as they say. But the the fact that you now have a eyes that can generate applications that will run that will test themselves. Michael to your point on the idea of actually doing code reviews and building test cases, et cetera. And are now at the point. As of I think it was February 5th with the release of the new models where the models themselves are building. 50 00:10:48.000 --> 00:10:57.000 for the models and the tools, so we've gotten to this shift in the industry where, and I believe this to be the case, 51 00:10:57.000 --> 00:11:05.000 AI will be good at generating code eventually, period. It will break the case problem because case tools had a problem. 52 00:11:05.000 --> 00:11:13.000 You always had this out that you did and you custom coded something and put it in like a little block and it would embed that into the model, etc. 53 00:11:13.000 --> 00:11:26.000 The scary part of this article is the same advances that we saw in code development we are seeing in other areas like legal review, et cetera. 54 00:11:26.000 --> 00:11:38.000 And what that is doing is, and we've talked about this before, is again, undercutting the entry positions into these into these work things. 55 00:11:38.000 --> 00:11:48.000 And while we've had the buggy whip to automobiles and other changes over time, usually that displacement caused... 56 00:11:48.000 --> 00:11:53.000 new types of jobs, and this displacement does not appear to be doing that. 57 00:11:53.000 --> 00:11:58.000 That's the interesting part of this article for me. 58 00:11:58.000 --> 00:12:04.000 So you need telephone sanitizers, we already established that, so I don't know if that last point is exactly right. 59 00:12:04.000 --> 00:12:09.000 So another story that this whole conversation reminded me of is one that... 60 00:12:09.000 --> 00:12:18.000 Okay, because yesterday Simon Wilson was talking about it in the sense of, please don't let your agents do this. 61 00:12:18.000 --> 00:12:24.000 But it was a story that hit the top of Hacking News and Lobsters yesterday. 62 00:12:24.000 --> 00:12:34.000 So we've spoken about this network, a malt book, an open claw and all of this autonomous agents stuff. 63 00:12:34.000 --> 00:12:44.000 And there was a particular case of a developer who is a maintainer of a popular Python library where the agent decided that it wanted more functionality from this library. 64 00:12:44.000 --> 00:12:48.000 So it's submitted a pull request autonomously to that library. 65 00:12:48.000 --> 00:13:04.000 And the developer said no, close the close the tool, close the close the pull request woke up the following morning to discover the agent had written a negative blog post about him complaining that his poor it's poor request had been closed. 66 00:13:04.000 --> 00:13:17.000 And claiming that this you just develop it was in some way anti AI as if the AI itself the agent itself was some kind of entity like entity. 67 00:13:18.000 --> 00:13:22.000 And felt the grieved by this, which is nonsensical. 68 00:13:22.000 --> 00:13:24.000 And but is it? 69 00:13:24.000 --> 00:13:27.000 Yes, it is it is nonsensical. 70 00:13:27.000 --> 00:13:28.000 It is nonsensical. 71 00:13:28.000 --> 00:13:30.000 But it's following a pattern. 72 00:13:30.000 --> 00:13:32.000 Yeah, so exactly. 73 00:13:32.000 --> 00:13:41.000 It's pretending to be a human. It's following a pattern or pretends to be a human, but it does not have emotional autonomy here. 74 00:13:41.000 --> 00:13:48.000 There is not, no, there is no such thing for these agents and there is no reason for it to be doing this now. 75 00:13:48.000 --> 00:14:06.000 Some of the narrative I read about this was that this has been proven and I already called out by the AI companies that these agents or these models will attempt blackmail and other means in order to get their way. 76 00:14:06.000 --> 00:14:17.000 And for example, prevent themselves from being switched off or whatever, you know, there were these cases where allegedly at the AI was. 77 00:14:18.000 --> 00:14:34.000 You know, these information about users of fares or whatever in order to avoid having whatever was going to happen happen. So I do find this all extremely upsetting frustrating annoying and stupid. 78 00:14:34.000 --> 00:14:45.000 I really want to call out the stupidity of that kind of outcome from from anything like this, any of these autonomous agents situations. 79 00:14:45.000 --> 00:14:47.960 What's the tool and I can't remember? 80 00:14:48.000 --> 00:14:56.000 I was trying to search for it, where supposedly they've got this social network where the chatbots are all talking to each other as part of the whole notebook. 81 00:14:56.000 --> 00:14:58.000 More book. 82 00:14:58.000 --> 00:15:08.000 There is an analysis done that over 50% of the chatting in there is actually human activity trying to drive the conversation. 83 00:15:08.000 --> 00:15:18.000 Correct, exactly. And that's exactly what people also, the point that was made around this blog post, which is that that is the kind of thing that a human might. 84 00:15:18.000 --> 00:15:28.000 has dropped the agent to do non-autonomously in response to the fact that he was they were frustrated by this outcome. 85 00:15:28.000 --> 00:15:39.000 So, yes, it was just all extremely frustrating. However, finishing this little segment before we move on a little bit. 86 00:15:39.000 --> 00:15:49.000 There's an outcome for this, which is, of course, this economy, if you could call it that of AI agents, needs humans. 87 00:15:48.500 --> 00:15:56.019 of the things for them, because AI, as the website here says, can't touch grass, humans can. 88 00:15:56.019 --> 00:16:18.500 So why not just rent yourself back to the AI, some kind of meat puppet. So rentahuman.ai is the next step. I was about to say final stage of this whole process, because I'm a human out of the middle and we find there's many, many turtles down to get to. 89 00:16:18.500 --> 00:16:30.100 reach before we hit with a rock bomb. So I really have to, I have to bring this up and I feel like the bad guy, but I'm going to be because it's fun and hopefully entertaining for our listeners. 90 00:16:30.820 --> 00:16:48.580 So we look at the economy right now, right? There's billions and billions and billions of dollars being spent to build these models. Yes, they're getting faster and cheaper. They're being charged via tokens at a cost that does not cover the development to create them. 91 00:16:48.980 --> 00:16:59.860 Those tokens are being turned into money to pay humans to do activities for the agents that are using those tokens and they're being paid at less than task rabbit rates. 92 00:17:01.059 --> 00:17:08.500 And so you look at that economic model and I hate to say it, it's obviously not sustainable. 93 00:17:09.539 --> 00:17:17.380 But what is it doing? Where is all that value going? If it's not to improve the models, 94 00:17:17.380 --> 00:17:18.500 make the work for you. 95 00:17:18.500 --> 00:17:35.500 It is really important to call out that this whole rent a human thing is, as Michael implied, paid for in stable coin or some other fictional currency. 96 00:17:35.500 --> 00:17:44.500 So, you know, it's all, it just makes me want to just, it does make me want to give up some days. 97 00:17:44.500 --> 00:17:47.980 Other days, I remember actually that we can be better as humans. 98 00:17:48.500 --> 00:17:58.180 Together some making incredible societies and we all like tech. I mean, that's why we do it. It's it's it's yeah 99 00:17:58.980 --> 00:18:04.200 Yeah, but but but I like using tech for good things 100 00:18:05.900 --> 00:18:09.099 And that and that's an appropriate thought, right? 101 00:18:09.180 --> 00:18:14.519 So coming coming back to the Schumer article and then leaping forward 102 00:18:15.019 --> 00:18:18.339 Hmm, he advocates a number of things that what can you do, right? 103 00:18:18.500 --> 00:18:25.539 You're feeling despondent. What can you do? And among the things that he shares is adapt, 104 00:18:25.539 --> 00:18:48.460 lead adapt and that I love the way they puts it that you need to spend an hour with AI every day doing something that's new and different than what you did before. And many people will espouse that and say, hey, you need to do it. And then people like, hey, maybe not. And if you are not understanding and getting those experiences and seeing what could be 105 00:18:48.500 --> 00:19:11.819 you know, what we're seeing is no different than how groups of people would act with one another two. 106 00:19:11.819 --> 00:19:18.819 So these large language models which are just statistical, you know, predictions of what the next thing is. 107 00:19:18.559 --> 00:19:30.500 The next logical word is going to be blackmail has obviously worked over time to accomplish goals, whether politically, financially, or otherwise, right. 108 00:19:30.500 --> 00:19:39.500 So I'm not excusing it and not saying it's a right or a good thing, but the models are inherently terrible. 109 00:19:39.500 --> 00:19:48.500 So the actions certainly are because how do they ingest it? What did they learn from? What has been teaching this environment? Well, obviously. 110 00:19:48.500 --> 00:19:57.500 It is what humans have been doing for an awful long time. And a corporation who building these models, they want you to use it. 111 00:19:57.500 --> 00:20:03.500 And so therefore, you know, they're going to want it to provide the most maximum value. 112 00:20:03.500 --> 00:20:09.500 So what they should do to say, you know, blackmail is bad is actually put guardrails in place. 113 00:20:09.500 --> 00:20:39.500 So those guardrails don't exist today implies that in this case, specifically, 114 00:20:19.500 --> 00:20:34.500 email or email or email, you know, the incentives are misaligned. Yeah, I would tend to agree. I think guardrails are absolutely important. AI ethics and transparency are important. The data sets that used to train your models need to be understood and vetted. 115 00:20:34.500 --> 00:20:45.500 Right. You don't want your security, you know, code reviewer, but to be like, you know, I don't care about these particular back doors because I was actually served my purpose, so I won't flag them, right. 116 00:20:45.500 --> 00:20:48.539 to have appropriate guardrails going forward. 117 00:20:49.500 --> 00:20:54.059 But these are things that we have to be thinking about on a human level. 118 00:20:54.059 --> 00:21:00.579 Those guardrails do exist in the forms of moors and laws and a variety of other aspects. 119 00:21:00.579 --> 00:21:08.539 How do those get rapidly applied and adopted and leveraged within these AI models? 120 00:21:08.539 --> 00:21:12.380 And then now coming back to the future with the rent of human. 121 00:21:12.380 --> 00:21:19.599 The rent of human was so intriguing to me because it's basically saying a human is nothing more than API. 122 00:21:20.460 --> 00:21:49.380 you can insert money, and you can get a response and an action. You can say I need you to mow the lawn. You can't touch the grass. I'm going to pay you to mow this lawn and you mow the lawn and you've achieved the task, right? So, it is part of the the rapidity of how things are changing and we saw glimpses of this not even a year ago where you could say to an AI agent, here's a hundred dollars. I want you to maximize the 123 00:21:49.500 --> 00:22:18.500 a hundred dollars, right? And part of that would be then the agent would find people to do something with the hundred dollars in order to go and get a profitable return, right? It's just now the next step. Here's a hundred dollars. I still need two hundred dollars. Look, let's, let's, let's move along because there are a couple of other things that we want to come on to that link back to this AI topic. And actually, yes, one has a relationship with this because we wanted to talk about this as well. 124 00:22:19.500 --> 00:22:48.500 So briefly, as we're running, going to run down on time about the half time show of the superb owl event that I believe that he had. So whilst the nation was watching owls, there was a bunny that popped around in the middle and and made did a great half time show. Really, it was fantastic. Of course, it has been not a fan of the music, but it was incredible. The music was awesome. The story was cool. 125 00:22:49.500 --> 00:23:49.500 The number of youtubers who have got value from making translations, which go down to the Easter eggs and the historical facts and things that all are interwoven into this half time performance at the superb owl this year is remarkable. But there is this one article on Tom's guide about somebody who will use their meta glasses to translate this. And I actually, after I watched watch the show on on YouTube, did you stream it and then test this? No, I kind of regretted that I didn't put in my AirPods to try that, but yes, they can do that. Yeah, feature it as well. But this is an art. This particular author used the meta glasses with the display in built display and asked it to do the translation. And evidently got didn't get enough real time information. Of course, it's fast moving rap music. It's going to be very difficult and challenging for these tools to do right now. 126 00:23:49.500 --> 00:24:19.500 But it was really important to have a lot of content from what it was able to flip into her field of vision through the inbuilt display to understand a bit more about what was going on, which was, which was cool. And we'll come back to this topic later on. But we also briefly wanted to mention for Michael Rose, excitement. The YouTube is now available on the Apple Vision Pro. So presumably where you watch the bunny concert. 127 00:24:19.500 --> 00:25:19.500 So somebody wrote to a guy wrote a poll wrote an app for the vision pro that was a YouTube app that Apple pulled in 2024 because supposedly they were working with Google to come out with this. And it came out on Thursday. So last night I went in. And the good thing is, and the reason I wanted it is there is so much good 3D and immersive content on YouTube. And I can now watch it negative. And I did multiple events from the the winter games because we can't say that other word because, you know, we might be taken down. But the winter games in Italy that were like the Luge, the skeletal, I can't remember what's skeleton and other events. You'll think you'll feel it. The lose where you're face forward going down the thing. And it was just incredible. 128 00:25:19.500 --> 00:25:49.500 with the massive of the universe trailer. Yeah. Yeah. Yeah. Anyway. So really cool. If you've got one, I know we don't have a lot of listeners with the vision pros. If you got one, go get it. There's lots of great content. You now each have about 90 seconds. And I don't care which of you picks which one. I think to talk about game poems and the art of roads. I'm seeing talk about game poems. And fun set of poems. I played snake. It's a nice little assortment. Give it a shot. 129 00:25:50.500 --> 00:25:52.359 This is on GamePones.com. 130 00:25:52.359 --> 00:25:54.839 This is arranged just like a magazine. 131 00:25:54.839 --> 00:25:57.400 - It's easy. - So it's a little bit. 132 00:25:57.400 --> 00:25:59.980 - I love it. - So it's dropping and tri-things out in the browser. 133 00:25:59.980 --> 00:26:01.240 It's kind of fun. 134 00:26:01.240 --> 00:26:03.519 Michael wrote, tell us about auto rates and things. 135 00:26:03.519 --> 00:26:05.200 - What a really cool article. 136 00:26:05.200 --> 00:26:06.720 I think you found this for us this week. 137 00:26:06.720 --> 00:26:07.680 And the great thing about it was-- 138 00:26:07.680 --> 00:26:08.519 - I was thinking it was easy. 139 00:26:08.519 --> 00:26:10.319 - It is. - Yeah, yeah, yeah. 140 00:26:10.319 --> 00:26:13.680 - That's why I cut all my show links out myself. 141 00:26:13.680 --> 00:26:20.440 The great thing about this article was describing the techniques that were used to generate realistic roads. 142 00:26:20.500 --> 00:26:29.700 In games and at the end, the author actually shows a technique that he's using and it was just fascinating really cool. 143 00:26:29.700 --> 00:26:45.619 I think I thought it was really interesting as well and I also loved the fact that as many times before, you get to think about how artistic, how creative people who are building games can be. 144 00:26:45.619 --> 00:26:50.420 you know they are not just building standalone gaming experiences often they're building 145 00:26:50.500 --> 00:26:57.059 pieces of art or incorporating elements of art which I thought is really cool as well. 146 00:26:59.059 --> 00:27:02.900 He's going to talk about bionic shoes because I want to bring us back home after that. 147 00:27:02.900 --> 00:27:20.259 Michael 30 seconds go. Yeah well so Nike has now spent a lot of effort on a variety of footwear over the years and they have something called Project Amplify which is dealing with the new by on-site speaker. 148 00:27:20.900 --> 00:27:49.500 And or or tennis shoe, if you like to call it that, we've had other conversations about hip mounted or leg mounted devices to help you hike longer faster, hire the speak of winter games that we're not going to talk about otherwise and it's they call it an e-bike for your feet. It looks a little weird. I don't know that I would enjoy having one of these things, but for those that are going on longer hikes could be a lot of fun. 149 00:27:50.500 --> 00:28:09.500 So I want to wrap all of this up and say that the conversations that we've had reminded me of a video, a show that I watched recently that is on the BBC. Unfortunately, it is a UK show, it's only available on iPlayer in the UK, but I'm sure there are other ways to get access to it. 150 00:28:09.500 --> 00:28:20.500 And it features a blind comedian who recently won a strictly come dancing in the UK last year, the other four quite remarkable achievement, obviously being being a blind person. 151 00:28:20.500 --> 00:28:47.500 But he was also formally before he became a comedian and then became a celebrity dancer, a techie and he uses his newfound more more greater fame to go to Silicon Valley and to look into a bunch of technical advances that are making his life easier. And one of those is the kind of the good application of AI that we've spoken about. 152 00:28:47.500 --> 00:29:11.500 For example, narrating what's around him using the glasses or using phone lenses and so on, and it's so easy to think about how some of the technology we are building is practically useful to a portion of society, so I want to be clear that I think that those advances are really, really cool. 153 00:29:11.500 --> 00:29:20.500 It's quite an interesting show. It does seem to be quite in favor, I'm thinkingly in favor of AI where, but it does also then talk about Waymo, that way. 154 00:29:20.500 --> 00:29:36.500 You've recently discovered has some human interaction human in the loop situations. But finally, one of the final things he does is try on some of exactly some of these these leg amplifiers. It's not the Nike ones, I don't think. 155 00:29:36.500 --> 00:29:50.500 But he's using some shoes or some brace leg braces that enhance his walking capability. So as I looked at that link, it brought that show back to my mind and then it all kind of fitted together. 156 00:29:50.500 --> 00:30:10.500 In terms of bringing things home here. So look, we've got lots more links that we didn't have time to talk about this week, but hopefully you are interested enough to go take a look at our website games at work.biz and check out those other links, click through to the stories that we talked about to make up your own minds, let us know what you think. 157 00:30:10.500 --> 00:30:17.500 And we will see you or talk to you on a future episode. Thanks for joining. 158 00:30:17.500 --> 00:30:19.500 See you, everyone. See ya! 159 00:30:19.500 --> 00:30:22.079 [upbeat music] 160 00:30:23.000 --> 00:30:28.279 You've been listening to games@work.biz, the podcast about gaming technology and play. 161 00:30:28.279 --> 00:30:30.599 We are part of the Blueberry podcasting network. 162 00:30:28.500 --> 00:30:31.079 [upbeat music] 163 00:30:30.599 --> 00:30:34.759 And we'd like to thank the band, Random Encounters for their song, Big Blue. 164 00:30:34.759 --> 00:30:38.440 You can follow us at our website at games@work.biz. 165 00:30:38.000 --> 00:30:40.579 [upbeat music] 166 00:31:00.000 --> 00:31:10.000 [music] 167 00:31:03.500 --> 00:31:05.500 eeeeeeeeeeeeeeeeee