ChatGPT considered
The enthusiasm for applications of generative AI reminds me a lot of our enthusiasm for web surfing in the 1990s. How many of us are still spending hours surfing the web?
Last Saturday I sent two newsletters at the same time. One was written by ChatGPT and one was written by yours truly. The topic of both newsletters was dual eSims. But the idea of sending both was to compare what the AI wrote on the topic (based on the instruction: “write about using dual e-sims in the style of stewart alsop”) and what I actually wrote.
ChatGPT did its usual remarkable job. It did summarize accurately the state of dual eSims, focusing mostly on how they help people travel internationally without paying roaming charges. What it wrote was true and written pretty well (although I could argue about how much it was written in my style).
But (you knew there was a but…) it reflects mostly what is already known about dual eSims. What I wanted to write about is whether the inclusion of dual eSims in the iPhone 14 Pro means something more, something that would be relevant to every mobile phone user. I don’t think an AI can speculate about a future that has no precedence. (One reader got the difference and wrote: “Carriers allowing a person to use his cell as a sort of an IP address is more what you allude to. The phone should just be like your IP but you should be able to have another way to call that would work across platforms like a URL that would be DNS’d to your number. Or something along those lines. We’re getting close to that being feasible technically and politically.” Thank you, Jose! I’ll come back to that but I’m talking about AGI (artificial general intelligence) right now.)
We are all fascinated by ChatGPT and Dalle and Lensa and the other applications of AGI that have made such a splash in the last few weeks. I know I am. But I also have this very strong intuition that this isn’t the point of these foundations. Indeed, I keep thinking that the parallel experience I had in my tech life was when I discovered internet surfing in 1994. I was the editor of InfoWorld then; I had to get our IT department to give me a copy of Mozilla on a floppy disk and explain what an IP address was. Once I got it installed, I spend hours surfing the internet. I was amazed by every time I clicked a link. Just like we are amazed today by ChatGPT et al.
But we didn’t keep surfing the internet; instead, the underlying technologies of the internet evolved and we started using it as a utility, a place to search, to learn, to inform (or misinform), and ultimately as a platform for applications that provided specific value: productivity, social media, advertising, marketing, and on and on.
I’m thinking that the same evolution will happen with AGI: It will disappear into the infrastructure and be applied to specific use cases. I have seven portfolio companies; I can tell you exactly how those companies can use this technology, by integrating it into what they do for their customers. (Dear CEOs: Pay attention. This will be a competitive issue in the next 12 months. Make sure you get there first!)
Bottom line: Enjoy the experience of being amazed by being able to write, make images, videos, and all the other stuff (even cheating on homework), particularly when it becomes multimodal. This is what makes technology so interesting and fun. But the real action will start when you begin seeing these capabilities built into the apps you use every day.
Hi Stewart,
Love it and agree!
Yet the way you're outlining it here you're talking about just integrating AI. What about reimagining what can be done from scratch with AI? Aren't those going to be the real killer apps? EG just as eCommerce and Amazon arose from TCPIP what will be the equivalent in AI that was simply not possible before? Any ideas? (I'm busy listening to people smarter than me before I offer my own)
Best, Michael