Illustration: Sarah Grillo/Axios
A new artificial intelligence thingamajig called ChatGPT has made waves on the internet this week.
Why it matters: ChatGPT, essentially an artificial intelligence (AI) interface that can text you like an all-knowing human, could herald major disruption ahead for big tech companies — search in particular.
How this works: Just type a hint into its interface, no different than the way you use Google – instead of returning links, ChatGPT will write back to you in paragraphs.
- When asked “how do you curb inflation”, ChatGPT gave an answer that Jerome Powell might find good:
- “There are several different ways to try to curb inflation, but most involve using monetary policy to manage the supply and demand of money in the economy,” the bot’s answer began.
Game status: Experts have hailed ChatGPT as a major breakthrough, developed by a company called OpenAI, which has been working for decades to create a bot that can text with humans like a real person.
- ChatGPT is working so well that people are starting to understand the power of chatbots, as tech writer Alex Kantrowitz told me Friday on the podcast What Next: TBD.
- “For years, we’ve been hearing from tech companies like Meta, Google, and even Amazon that voice and chat talking to computers is the future — and the actual product has never really been measured,” he said. “Now we’re starting to see that there’s actually powerful technology here.”
- Chat GPT is A step further than the clunky text interactions you might have with your bank or airline.
- In the first five days, more than 1 million users signed up to try ChatGPT, according to the president of OpenAI Greg Brockman.
🍪 Plot: When asked how to make chocolate chip cookies, ChatGPT gave me a short, clear recipe that seemed pretty standard. Compare that to Google Search, which returns links to lengthy, overwhelmed blog posts that require endless scrolling to get recipes.
- These recipe sites are part of an entire business ecosystem built around optimizing for Google search — including financial sites like Investopedia, media outlets, and other services.
- That’s part of the reason Google is investing heavily in chat AI. The tech giant made $149 billion from its search business last year; more than half of its total revenue.
Yes, but: ChatGPT doesn’t know if what it says is true.
Between the lines: Writers and teachers are already concerned about AI’s ability to write papers.
- Paul Krugman of The New York Times asked if ChatGPT was a good fit for a tech job—he was one of several writers who wrote a column using the interface in the past week.
- Axios’ Ina Fried called it “terribly good,” but was also concerned about the fact that ChatGPT didn’t say where it got its information from and could be “definitely wrong.”
Zoom out: It’s worth noting that this new AI was released not by a major tech company, but by a risky startup. OpenAI made waves earlier this year when it released Dall-E2, an AI that can be used to create artistic images. Artists freaked out.
- It makes sense that a small, private player is the one to make waves with AI; these companies can release something experimental with less risk of angering users, shareholders, or advertisers if something goes wrong.
- Open AI does have some powerful friends in the industry, though. It announced a partnership with Microsoft in 2019.
Reality check: The future is not here yet. Users have spotted flaws — like having a robot tell you how to shoplift.
- And, there are some big red flags to watch out for: AI tends to reflect human bias and manipulation (like when a Microsoft chatbot started cheering Hitler). So far, ChatGPT seems to be doing a better job of avoiding this.
What’s next: Expect to see more bots from bigger players. “It’s playtime at Google,” Kantrowitz said. “I don’t think it can stand by for too long.”