Most of what the developer platform initially enabled were silly games. But those games expanded the range of what the social network was good for. They gave Facebook a built-in advantage over competing sites and made the platform more addictive. In 2008, Facebook overtook its rival MySpace in unique visitors, and by 2009, time spent on the site had skyrocketed. This was still in the early days, back when Facebook was a “social-network site” rather than part of the now-ubiquitous “social media.”
FarmVille tapped into Facebook’s ethos of networked participation and fit Facebook’s algorithmic News Feed like a glove. Zynga, the game’s developer, reached a multibillion-dollar valuation based entirely on its ability to create games within platforms that people could not stop playing.
A little more than a year in, it seems as though devout ChatGPT users generally fall into one of four buckets. First, there is the subset who were already using machine learning in their jobs. Generative-AI products such as ChatGPT do the job better than the older products that came out of the Big Data hype cycle a decade ago. These are the people who remain the most genuinely excited about the revolutionary potential of the technology. But their group is also pretty narrow.
Then there are students using ChatGPT to cheat on tests or for help with homework. ChatGPT usage declined in the summer, when kids were out of school, surprising approximately no one. There are also the managers who spent the past year or so using generative AI to cut costs while sacrificing quality, particularly in the media industry. Finally, there are the hobbyists and hype bros, people who just like to play around with ChatGPT because it feels like the future to them, or because they think they smell profit. The same type of Twitter accounts that spent 2021 hawking NFTs—and 2022 crowing about the metaverse—are now all-in on prompt engineering.
These were the transition years, when niche social networking expanded into social media. Facebook didn’t make FarmVille: Rather, during this crucial chunk of time, FarmVille made Facebook. OpenAI is at a similar juncture today. 2023 was supposed to be the year of the ChatGPT revolution. The AI doomers and AI bloomers were everywhere, yet viewed with a bit of critical distance, the technology is still little more than a toy.
OpenAI is inviting developers in 2024 to build tools that could themselves become lawsuit targets. OpenAI has created a moderation system to weed out GPTs that violate its brand guidelines and usage policies, but there is a pretty wide gap between OpenAI’s position on acceptable GPT uses and the position of potential litigants.
The immediate future of generative AI looks a bit like Facebook’s past.
By David Karpf
Illustration by The Atlantic
Listen to this article
00:00
10:11
Produced by ElevenLabs and News Over Audio (NOA) using AI narration.
Updated at 2:49 p.m. ET on January 17, 2024
ChatGPT has certainly captured the world’s imagination since its release at the end of 2022. But in day-to-day life, it is still a relatively niche product—a curiosity that leads people to ask questions that begin “Have you tried … ?” or “What do you think about … ?” Its maker, OpenAI, has a much more expansive vision. Its aim is seemingly to completely remake how people use the internet.
For that to happen, the bot needs to be more than a conversation starter: It has to be a functioning business. The company’s launch of the new GPT Store on Wednesday was an ambitious step in that direction. Initially announced two months ago, the GPT Store allows the product’s business and “ChatGPT Plus” users—those paying $20 a month for an upgraded version of the service—to create, share, and interact with customized AI agents (called GPTs) that are tailored to specific tasks. The company claims that its users have built more than 3 million of these custom bots since they were granted the ability to do so in November, in preparation for this launch.