Derek unpacks his thoughts about GPT-4 and what it means to be, possibly, at the dawn of a sea change in technology. Then, he talks to Charlie Warzel, staff writer at The Atlantic, about what GPT-4 is capable of, the most interesting ways people are using it, how it could change the way we work, and why some people think it will bring about the apocalypse.
If you have questions, observations, or ideas for future episodes, email us at PlainEnglish@Spotify.com. You can find us on TikTok at www.tiktok.com/@plainenglish_
In the following excerpt, Derek talks to Charlie Warzel about the untapped possibilities of AI in everyday life.
Derek Thompson: So I wanted to bring you on because I consider you a bard of uncertainty when it comes to technology. You are very, very good at diving into deep, murky waters all the way to the bottom, seeing what’s going on at the ocean floor, and coming back to the surface and being like, “Holy shit. There’s some weird shit down there.” You’re very good at explaining the quality of weirdness that you observe in all of these spooky corners of technology. And AI, I consider a very spooky corner of technology.
So I do think that lots of conversations about AI that I hear on other podcasts can immediately go into the stratosphere of speculation very, very quickly. And the truth is, we are headed toward the stratosphere of speculation in just a few minutes. But before we hit blastoff, I want to start by anchoring the conversation to things that are actually happening, actual news. GPT-4 is out, the fourth generation of this technology from OpenAI. I am using it. I have forked over the $20 a month to get access to the ChatGPT that’s powered by this tech.
Charlie Warzel: So I think right now there’s many different camps. There’s the true sickos, like you and I, who have forked over the money because we just need to experience this now and get our bearings, and we’re going to write about it. There’s that kind of exploratory crew. Then there’s the people who may or may not know that they’re using it. So that’s the people who want to use Bing’s new chatbot. That is infused with, as we now know—it was speculated, but we now know—GPT-4 or an early version of GPT-4. There’s all these different, think of it as software updates. And so in that sense, Microsoft just rolled that out to anyone who wants to use it. So you could say that millions of people are using that today to do that prosaic search chatbot thing that we’ve been talking about for a while now.
Then there’s the sort of enterprise group, which I think is a super-fascinating use case. This is how Uncle Steve or Aunt Molly or whatever is going to start to encounter the technology. And that’s these partnerships through OpenAI, which has a partnership with the consultancy firm Bain to work with clients like Coca-Cola, these big, huge companies. They release an API integration, which is essentially allowing different programs to access the tool or different developers. And so we’re seeing Slack is developing one of those to respond to messages or summarize big, long threads in very concise bits. Salesforce has that for their customer management stuff. They’re going to roll that out, and Salesforce is used by tens of millions of people to do really boring stuff across businesses everywhere.
And then you have the announcements this week from both Google and Microsoft that they’re going to put this generative AI tool stuff inside all their workplace clients. So that’s Docs, Calendar, Gmail, Slides, whatever. And that’s going to be able to do and automate a bunch of that different stuff, in the way that you currently have autocorrect for your Gmail. So really, it’s hard to know how many people are using this tool and in what way it’s the purest version, which is the “I pay $20 a month and I’m just going to experiment my face off” version. But I do think that there’s a number of people who are encountering this in a very organic way just through their jobs, or at least will very soon.
Thompson: Charlie, there’s all these ways that people are using this and showing off their usage online on Twitter. Give me an example of what you consider one of the most clever applications of this technology.
Warzel: There’s one that I saw yesterday, and my journalist brain is like, “I love stunt journalism.” And I was like, “This is perfect.” I think you probably saw it somewhere along the line if you’ve been looking at this stuff. But somebody basically said, “I want to take $100 and start a business, and I want to have GPT-4 make decisions for me to try to turn that into as much money as possible without doing anything illegal, and just refine the steps along the way.” So it’s like, what kind of business? And I think they decided on environmentally friendly products, like silverware and weird stuff like that for camping. But then, OK, what will the website look like? What will the logo look like? And then feed that into a stable diffusion or Midjourney prompt and get something out and refine it.
And it’s really fascinating. It’s really cool to see. I think we’re used to—and I noticed this with Bing search—we’re so used to being like, “Give me one discrete answer,” and not having the computer or the machine reason at all or make multiple inferences. But the genius of AI-assisted search is you can say, “How do I get this IKEA bed to fit?” Or “Can I get this IKEA bed to fit in the back of my Ford Fiesta?” And it will go and do all the different calculations and look up all the different stuff. So I think that’s where I and a lot of normies need to start to change our brains. It’s like, how can we get this thing to start thinking a little bit on our behalf, or at least taking steps and making connections. Because that’s what this technology ultimately does. It just makes lots of inferences, right or wrong. As opposed to need an answer, ask for an answer, get an answer, transaction over.
This excerpt was edited for clarity. Listen to the rest of the episode here and follow the Plain English feed on Spotify.
Host: Derek Thompson
Guest: Charlie Warzel
Producer: Devon Manze