This week’s podcast is about AutoGPT / AgentGPT and two other cool new technologies.
You can listen to this podcast here, which has the slides and graphics mentioned. Also available at iTunes and Google Podcasts.
Here is the link to the TechMoat Consulting.
Here is the link to the China Tech Tour.
Here are the three technologies mentioned:
- 5.5G and immersive and experiential video everywhere.
- AutoGPT and AgentGPT.
- Human organizational structure adopting software architecture.
- The Winners and Losers in ChatGPT (Tech Strategy – Daily Article)
- Why ChatGPT and Generative AI Are a Mortal Threat to Disney, Netflix and Most Hollywood Studios (Tech Strategy – Podcast 150)
From the Concept Library, concepts for this article are:
- GPT and Generative AI: AutoGPT
From the Company Library, companies for this article are:
- OpenAI / ChatGPT
Photo by Andrew Neel on Unsplash
Episode 162 – Autogpt.1.transcribe
Fri, May 05, 2023 10:30AM • 28:50
gpt, video, create, towson, auto, china, prompt, task, give, agent, digital strategy, software, starting, put, ai, companies, topics, number, called, jeffrey
Welcome, welcome, everybody. My name is Jeff Towson, and this is the tech strategy podcast where we analyze the best digital businesses of the US, China and Asia. And the topic for today, auto GPT, and other tech that I am just super excited about really three. But the big one is auto GPT, which is, I mean, it’s just the last couple of weeks this thing has emerged. It’s stunning. It’s called God Mode, AI. It’s called agent GPT. It’s if you haven’t heard about it, it’s pretty unbelievable. And I’ll give you the basics of what it does. And the short version is it’s AI that can prompt itself. Instead of GPT chat GPT where you give it a prompt to do something, you don’t give it a prompt to give it a goal, and then it creates its own prompts and goes after that goal. And that can be everything from create a website, create an app to go get me a pizza, and have it delivered to my home. Right, I’ll give you something that’s a real one you can do. Anyway, so we’ll go into that. And then there’s two other sort of emerging tech, which are I want to talk about, but that’s the big one auto GBD. And, okay, standard disclaimer, nothing in this podcast or in my writing or website is investment advice, the numbers and information from me and any guests may be incorrect, the views and opinions expressed may no longer be relevant or accurate. Overall, investing is risky. This is a non investment, legal or tax advice, do your own research. And with that, let me get into the topic. Now I don’t really have a concept for today, there’s no digital strategy, you know, conceptual ideas, economics, that sort of stuff. But we could put this under the topic of generative AI of which GPT is a type of we’re now on GPT. For, you know, that’s a type of generative AI, you know, there’s other ones. And then I guess we could put auto GPT as a subcategory there. So I’ll probably put those in the concept library, but it’s not really a major concept. It’s just a tech. But it’s important. Companies are open AI, which is behind all this, which is really, you know, tied with Microsoft. And then aging GPT, which is I’m not even sure it’s a company, like a lot of the cool stuff that’s emerging, it’s, it’s hard to tell if it’s even a company like mid journey, which I totally love. I use that all the time just for fun. I mean, I think that’s just 11 people just doing it. And you know, it’s up on Discord or whatever. But I’m not even sure what kind of company set up, if any yet. So anyways, we’ll call those companies but kind of not really at this point. All right, so let’s get into let me do an easy one. First, we’ll get to GPT auto GPT next, but three sort of new tech emerging phenomenon that I’m like really excited about. The first one would be just 5.5 G for mobile connectivity. I mean, everyone’s kind of knows 5g. Now, that’s rolling out. But 5.5 is right on the horizon. I mean, it’s coming really quite fast. You know, because it’s kind of an extension of five obviously. Now, there’s various use cases, I’m not going to go into those. That’s not really what I think is exciting. What excites me about 5.5 G is what they’re calling immersive and experiential video, also referred to as video everywhere. Like, okay, you look at 5.5 It’s sort of the new enhancement and expansion of 5g, it’s basically gets you a 10 fold increase in network performance over 5g. So five times higher speeds. So if you’re a mobile user, or a broadband user in your home, you know, it just a massive jump forward. 10 times more connectivity. So you know, you can connect more devices, IoT, that’s a big deal. It’s also more energy efficient. And what there’s a lot of uses for this people are talking about, oh, autonomous vehicles and industry transformation and IoT devices everywhere. Fine, fine, fine. That’s not really the one I want to talk about. The one that I think is just cool, is when you have that much connectivity and speed, you can put video into two way video into basically everything. You know, so you could think Metaverse, but you could think augmented reality you could you can basically put video in every hospital and every store and every street and you can just pipe high quality video in one direction. And then you can respond to it and the other so it’s interactive or immersive video everywhere. And that’s when sort of day to day reality starts to look kind of different. I mean, people talk about Metaverse, fine, that’s interesting. But you don’t have to go to that level of strapping an iPod onto your face. I mean, it can just be sort of glasses free, augment, you know, augmented reality, virtual reality 24k, streaming everywhere screens everywhere that you can interact with all the time. And that really becomes really interesting for b2c scenarios. And I was talking with aguaje, about this, like six months ago, and kind of looked at how they were thinking about it. And they had some strange terms for it, I don’t think they’ve quite got their thinking set on this. But it’s just like video everywhere, fully immersive. It’s basically when the physical world will just become intermeshed, with the online world, because their screens everywhere, and you can interact with them. Like it’s nothing. And so that’s kind of cool. Basically, 10 gigabit per second, no latency, real time interactive services. So the businesses you could build on kind of looking at what businesses you can build, when you can basically build on that is all types of digital and interactive services. And you can just kind of put them anywhere. You can offer people various services as they walk down the street, or through the mall or through the shops, or in the hospital, or in the doctor’s office, or whatever. And those sorts of interactive immersive experiences, facilitated by video, basically just become services. So that that was kind of just like, fascinating, and I’m not totally sure what’s leading, but that’s number one on my list. Let me get to a second short one. Well, let’s get into auto GPT. Okay, now let’s get to the big one, keep an eye out for that. I’ll see if they come up with some better language to describe that. But I know they’re prototyping a bunch of stuff right now. Alright, so let’s talk about auto GPT. Now, you know, generative AI fine, I think people get a sense of what that is, there’s techspace AI chat. GPT is obviously the sort of mothership there. Video runway is a really cool site that’s making videos and you can basically, it’s not to the point where you can just type in some text and create a video because there’s not enough information for there for what to do. But you can take a phone, and, you know, take a quick video, pulling a little toy truck across your desk. And if you give it that video, and then give it a style, it can turn that into a race car or whatever. So if you haven’t seen that go over to runway. It’s really cool. The videos they’re doing. Mid journey, obviously is the one I like I use it all the time to create no characters, pictures, videos. If you look at some I’ve been putting up some short videos recently. And if you look at those, there’s the main one I’m using. I’m in basically an office talking to a camera. The office isn’t real. It’s just me, the offices is mid journey created. And in fact, a lot of what’s in those videos is not real. I’m just creating backgrounds and whenever in mid journey, and you really can’t tell the difference anymore. That one’s pretty cool. Khyber where you can create music videos anyways, I’ve talked about these before. Okay, so why is auto GPT? So cool? Well, it’s basically the best way I’ve heard it described was it goes from search to answers to goals, which is okay, we all used Google for 1520 years. That searching searching for information, pretty cool. Then chat GPT jumps out, and we all start going from search to answers. Where I’m not searching, I’m saying give me the answer. And that was pretty awesome. And we’re, you know, that’s kind of a big deal. Now, that lasted all the four months. And now we’re moving from answers to goals, which is, look, I’m gonna, I’m gonna give you a goal, the AI, do it. And I’ll give you some examples of that. But basically what the AI does, it will take the goal that you give it and it will say okay, to accomplish this goal, we need to do four tasks. And then it will investigate each of those tasks and come up with an answer and then it will give you the goal. And that can be the answer to a question. It could be create something for me it could be write an app, it could be ordered me and deliver a pizza. So it’s sort of an IT can basically prompt itself and if the answer is not right, it will prompt itself again, so you give it a goal, it can run 100 prompts on itself to accomplish that goal and refine A goal and break it into subcomponents. And this is where it’s starting to look like not sentient, but it’s able to do things. And the two that came out, so that people call it like godmode AI. But the two versions that sort of jumped out at this, there’s auto GPT, there’s another one called Baby AGI. It’s starting to look like you’re creating and deploying an agent that can do things for you online, it can search, it can gather information, it can summarize, it can create, but increasingly can kind of do it in the real world. So I’ll give you some examples, sort of a so you could say, here’s some of the use cases we’re seeing, you can tell it to write code. So you you say write me, Python code that does this, you give it a goal, it will create it, it will test it, it will fix the errors, it will revise it, it can prompt itself hundreds of times, if you want it to create a mobile app for me, create a web page for me can do all of that. The one I gave the example I gave you is ordering the pizza, this is real, you can basically say, I want you to find the best pizza place near me order a large pepperoni with extra cheese and pay with a credit card. So then the auto GPT will because it has internet access, it will search the area, it will create the text, it will interact with the web pages. And it basically breaks it into a couple of tasks. That’s kind of the interesting thing is it tells you how it’s breaking it down. So the first tasks would be to search for pizza places near your location and compare the ratings prices and reviews. task one, task two, select the best pizza place based on your preferences and availability task three, call or visit the website of the pizza place and place the order using natural language. Next task confirm the order details delivery address and payment method. Bam. And it can do it. I mean, it’s it’s and you can even apparently, if there’s any issues that come up like order problems cancellation, modifying the order tracking the delivery, contacting customer service, it can basically deal with this too, depending how you know complicated you’ve done it. Yeah, that’s, that’s doable. Another example research. I gave it a couple of things to research for me. Here’s what I told it to do. I typed in as this is Agent, Ansari, Agent GPT, and auto GPT auto Jeep, and T is the tech agent. GPT is basically a website that draws on that. So you can go right now to agent GPT. And it basically goes into your browser. And that, you know, same way chat GPT is an interface for GPT agent. GPT is an interface for auto GPT, basically. So you can go to agent GPT right now and try this. And so I put in a couple goals about creating content. So here’s what I wrote, I wrote to create a headline and outline for a very popular five minute YouTube video for Jeffrey Towson, based on his writing about digital strategy. That’s what I gave it to do. It broke that into 1233 tasks. Task number one, which it did. Analyze Jeffrey Towson is writing on digital strategy to identify the most popular and engaging topics. This is what it came up with. I didn’t tell it to do that. Task number two, research and analyze successful viral YouTube videos to identify patterns and techniques that can be applied to create a viral video for Jeffrey Towson. Task number three, generate multiple headlines and outlines based on the analysis and present them to Jeffrey Towson for selection. That’s what it did. And then it gave me the result mean it just told me what it was doing. Here’s what it sent me it said after analyzing Jeff 1000s writing on digital strategy, the most popular and engaging topics include and it gave me five number one the rise of digital platforms in China’s economy. Number two strategies for companies to succeed in China’s digital marketplace. headline number three the impact of digital disruption on traditional industries. Number four, the future of E commerce and retail in China. Number five, the importance of data and analytics and digital strategy. Okay, then it goes on based on these topics a potential headline and outline for a viral 5g that YouTube video for Jeff Towson could be, quote, five digital strategy secrets to succeed in China’s booming economy, unquote. That’s pretty good. And when I’m looking at sort of the topics, it really did sort of go through everything I have on my webpage. Well, a lot of it’s paywalls. But those are fairly good summaries of what I write about, like, very thoughtful. So that was the headline it said, which I should do and then it gave me an outline for that, that one, article or video five digital strategy secrets to succeed in China’s booming economy. And here’s the outline. First point introduction to Jeff Towson, and his expertise in digital strategy point to the rise of digital platforms and China’s economy and the opportunities they present. Point number three strategies for companies to succeed in China’s unique digital marketplace. Strategy number four, the impact of digital disruption on traditional industries. And okay, it’s kind of conclusion Bababa and takeaways for businesses looking to succeed in China’s digital landscape. That’s a pretty good video. And I could mean I could take it further, I could say now write the text. Now write each section for me. So that’s apparently how you do research and writing now. And it’s, that’s pretty awesome. Here’s some other use cases. As writing articles, basically, you can create that and I can do other stuff. I gave it some other little questions here for fun. I said, come up with five keywords that will get my articles. I’m paraphrasing, come up with five key words I should put in my article so that I rank higher in Google. And here’s what it came back with. After analyzing the data, the five top five trending keywords for Jeffrey Towson, his target audience are number one, digital transformation. Number two, innovation number three, China market number four entrepreneurship, number five venture capital. So it basically scanned the content on my web page, which is open for reading, not the paywall, then it scan trending topics on Google and other places, and then it came up with keywords. That’s pretty great. It’s really going to change how you do stuff if you create this stuff. Let’s see other stuff. So other use cases. So research is one writing is one preparing podcasts. I could basically take this podcast, feed it in and have it cut it up and reorganize it and do other things. And you can do the same thing for videos. Because if you’re a video editor, most people use Adobe, well, Adobe now has AI in there. So I could take a video, like the one I just did with, you know, the fake background, which is mid journey, I could upload that video. And I could just tell it to do everything, make this more attractive as a video by putting in B roll which is like pictures of scenes and things, soundtrack music and text on the screen. And he could just edit the video on its own. It would find pictures that are compelling it would find background video B roll it would find sounds it would find music. And it can do the whole thing. And you can see videos of people doing this online where you just feed your core video in which has your content and then you just tell it to start editing the video. It’s pretty amazing to watch it do it. And let’s see other use cases philosophy, you can ask questions and there is one that’s floating around because people are this way, where there’s something called Chaos GPT, where they basically gave GPT the directive to destroy humanity, which I’m not kidding, destroy humanity. And how would it do it? And I don’t know. I’m not sure how far that’s gotten. But someone created it. You can read articles, but look up chaos GPT you’ll see it anyways. So this is all sort of pretty stunning. And this is why people are starting to think that like, you know, the whole like, is there a singularity is AI going to take off on its own? And it’s like we’re really close to that. I mean not I shouldn’t say that. I don’t know enough about the technology to know that. It’s starting to look like independent agents like because we can now create an agent that does what we want and we tell it what to do. And it can tell itself what to do you know gives itself the next steps, it gives itself the next task and then it prompts itself and does it. Now apparently GPT because the source code is proprietary GPT can’t edit its own source code. That could be a singularity moment when GPT can not only do tasks, create code to create websites, but he could change its own code. Can’t do that yet. I guess it’s not allowed. But I think the Terminator movies kind of showed definitively that that doesn’t hold. So anyways, what else do I want to talk about? I think that’s kind of it. It’s, it’s a big deal, you should go on and look at it. It’s pretty exciting stuff. And oh, there’s one last point of this is another idea floating around last point on this, that you can go from auto GPT to what they’re calling perpetual GPT. Or at least a couple of people on Tiktok are calling it perpetual GPT. Which is, Okay, you go from search to answers. Now we go from answers to goals. The next thing after goals would be systems, where there’s only one cost involved in any of this, which is the cost of the prompt the cost of the computing power to do this stuff, which is like 30 cents per prompt, something like that right now. Okay, so you can’t just prompt prompt prompt, indefinitely, unless, what you’re telling it to do, makes money. So if you say I want you to create a website that sells shoes, and I want you to take 20% of the revenue that you generate from that website and use that to run yourself, at that point, the auto GPT becomes self sustaining perpetual GPT. And it just make some money and then it reinvest it and it becomes a system that can kind of run on its own. And man can go on forever. So that’s the other ideas, you go from goals to systems, which means going from auto GPD, to perpetual GPT. And this isn’t just terms people are throwing around. But that’s floating out there. Okay, I think that’s most of what I wanted to talk about. This is going to be a short one today. All right, last one. So I gave you two ideas, the sort of video everywhere enabled by 5.5, G, auto GPT. The other one I’ve been thinking a lot about and I’ve been kind of listening to podcasts about this is sort of human organizations starting to become software architecture. And it came from this podcast I was listened to, to about someone who talked about software architecture. Now software architectures, and this is an old subject that’s been going on for 3040 years. How do you build software to accomplish whatever you need in life, and there’s always a bit of a conflict at the center of software architecture, which is you want to you have to create the mechanics of the system, if A happens, then B happens if C happens. So it’s got to be internally consistent. You set up the rules. But at the same time, the external environment in which any software lives is always changing. So if you want it to be rigid to some degree, so it’s stable and functioning. But if it’s too rigid, it becomes obsolete. And this is, you know, software, people talk about this, we have to write some code that works, let’s say for our ERP system, but we also know it’s going to have to adapt over time, because people are going to create new capabilities and the environment is going to change. So it’s got to be sort of rigid and solid at the core. But it also needs to be adaptable. And how do you reconcile those two things? So one of the answers software people talk about is modularity, we got to make things that are modular so that we can sort of stack them like bricks, and the modules all talk to each other, and we can build upon them and change them. And that’s sort of one solution to that. But you don’t want your entire architecture to be modular, you want parts of it not to be very modular, and you want other parts to be changeable. Right? So that it’s like an old subject software, people have been talking about this for years, decades, really. Okay. But then there’s this other idea of as business is increasingly being done not just by human agents, but also by software agents, which is kind of the auto GBD thing. You know, the software will act as an agent on its own, but the human will also sometimes act as an agent on their own. And all an organization is a company is is an association of people. That’s what an organ is. It’s an org chart. So we’re starting to see Human organizations look one, they’re having to incorporate more and more software. But also, they’re starting to look a lot like software architecture. So you know, Jeff Bezos famously wrote his I forget what year was 2002 2003 memo, where he basically argued that every division of Amazon has to be able to interact with all the other divisions, basically a modular structure. And there have to be clear interfaces between the modules. And that’s a very different structure than a hierarchical structure. And when you talk about agile companies, like let’s say, Alibaba, everything is based on teams that you deploy, and that move around Well, that’s very different than your traditional hierarchy with the boss and the vice presidents and so on. So it’s a type of, you know, this whole team’s approach is a lot like modular structure. So it’s kind of a couple of things happening. It’s like one, the workflows within a company are becoming increasingly software driven, because we use it for everything. So companies are starting to look like software architecture, but then at the same time, we have these new digital agents that are becoming super powerful, more working with them. So they have to work that way. So this is interesting thing where human organizations and software look like more and more the same thing. And the way we organize people, and the way we do software architecture, increasingly looks like the same thing. And that’s kind of this big, cool subject. And within that structure, the same, it’s the key question is still the same. How do you build a foundation that is solid, but also can adapt at the same time, if it if it can adapt, it will become obsolete, it won’t adapt to changing environment? If it’s too adaptable, there’s not enough sort of foundation for this thing. So it’s that same software question from 1980 is now being applied to humans Plus software. And it was I’ve been thinking a lot about that. I think it’s really cool. I don’t know where it’s going. But if you’re analyzing companies, I think that’s going to become a bigger and bigger part of everything. Okay, I think that’s it for content for today. Go over to agent GPT and start playing like, it’s really, really cool. And also go to read journey and runway. Those are the kind of the three I play with the most and then chat GPD, of course. And that is it for the content for today. As for me, I was in China for a couple of days in Shanghai, which was spectacular. Just a lot of fun. I mean, there were, there were Tesla’s and BYD s which is you know, Chinese brand for their the other major electric vehicle maker, let’s say top three. They were everywhere, like all over the city. And the funny thing is the BYD basically are copies of Tesla, they look exactly the same. You can’t tell them apart at a distance at all. Really a lot going on fantastic fun. And of course, went out and had some really good food, which is my standard first thing when I get back in China, I was actually supposed to go to a smart mine. I was supposed to go to a mining complex in sort of Northwest China this week, and then they’ve automated the mind. So you can do remote control and things like that you can have basically your miners on the surface and they control the devices. I was supposed to do that I got cancelled at the very last minute, which was a little bit frustrating. So I was able to do that again. Some time. I was really looking forward to it. I was reading all about autonomous mining rigs. So we’ll see. Yeah, but that’s it. Anyways, I hope this is helpful. And I think that is it for this week. Take care and I’ll talk to you next week. Bye bye.
I write, speak and consult about how to win (and not lose) in digital strategy and transformation.
I am the founder of TechMoat Consulting, a boutique consulting firm that helps retailers, brands, and technology companies exploit digital change to grow faster, innovate better and build digital moats. Get in touch here.
My book series Moats and Marathons is one-of-a-kind framework for building and measuring competitive advantages in digital businesses.
Note: This content (articles, podcasts, website info) is not investment advice. The information and opinions from me and any guests may be incorrect. The numbers and information may be wrong. The views expressed may no longer be relevant or accurate. Investing is risky. Do your own research.