ChatGPT just turned one, and even some OpenAI employees can't believe the way its first year has played out

• ChatGPT turned one today.
• When it was first launched, OpenAI leaders framed it as a "low-key research preview."

ChatGPT turned one today, and if you're surprised by its impact, you're not alone — so are some OpenAI employees.

When OpenAI was preparing to launch the first public version of ChatGPT, company leaders presented it as a "low-key research preview" — so low-key in fact, that some employees were unaware that ChatGPT had even launched, according to the Atlantic.

In the year since ChatGPT's launch, it has emerged as the poster child of the AI revolution. Over 100 million users accessed ChatGPT's website within two months of its launch, according to data analytics firm Similarweb. Analysts at investment bank UBS wrote in February that they "cannot recall a faster ramp in a consumer internet app."

ChatGPT's runaway success also kicked off an AI arms race among tech giants, inspiring a string of competitors from Google's Bard to Meta and Amazon's Q. And it's not just tech giants that are getting into the field: Funding for AI startups has defied the general decline in startup funding, according to a Bloomberg report in October.

But ChatGPT's reach extended far beyond tech. Its rising popularity stoked concern over AI replacing workers — and sparked an endless flurry of think pieces on when or if AI would ever do that. A Goldman Sachs report published in March projected that AI could disrupt as many as 300 million jobs worldwide.

At the same time, OpenAI has drawn criticism over its impact on online misinformation, its unauthorized collection of data to train its AI, and its environmental impact. The AI boom that ChatGPT started will likely influence how we browse the internet, the shape of copyright law, and more.

All that in the span of 12 months.

Today, the term "low-key research preview" has become something of a meme among OpenAI's employees. OpenAI employees have even used the phrase as laptop stickers, per the Atlantic.

OpenAI cofounder Greg Brockman used the phrase in a post on X on Thursday, writing that the team expected the "real moment of excitement" to come when they launched GPT-4. The first version of ChatGPT was powered by GPT-3.5, and GPT-4 was launched in March.

William Fedus, an OpenAI employee, also wrote in a post on X on Thursday that they didn't expect the first version of ChatGPT to make the splash that it did. He said that just a week before launch, Meta's Galactica bot was swiftly taken offline after churning out inaccurate information.

"The outside world probably thinks we're all joking, but this actually was our 'low-key research preview,'" wrote Fedus.