VANCOUVER, CANADA - 

This article has been adapted from the original version, which can be found on Gary Marcus' Substack.

As 2022 drew to a close, OpenAI released an automatic writing system called ChatGPT that rapidly became an Internet sensation: Less than two weeks after its release, more than a million people had already signed up to try it online. As every reader surely knows by now, the user types in text and immediately gets back paragraphs and paragraphs of uncannily human-like writing, stories, poems and more. Some of what it writes is so good that some people are using it to pick up dates on Tinder (“Do you mind if I take a seat? Because watching you do those hip thrusts is making my legs feel a little weak.”). Others - to the considerable consternation of educators everywhere - are using it to write term papers. Still others are using it to try to reinvent search engines. One rarely sees anything with this much buzz.

Still, one should not be entirely impressed, and there are a few important reasons for this.

Although ChatGPT can write about anything, it is also easily confused. As I told New York Times columnist Farhad Manjoo, ChatGPT - like earlier, related systems - is “still not reliable, still does not understand the physical world, still does not understand the psychological world, and still hallucinates.”

In other words, it makes stuff up, and on a regular basis. A fair bit of what it says is simply not true.

An example of this is ChatGPT’s claim that