There’s an appealing metaphor for ChatGPT running around on social media: It's a calculator for writing.
“I don’t know how to do it well yet, but I want AI chatbots to become like calculators for writing,” says Kelly Gibson, a high school teacher that was frightened at first but is now amazed at the possibilities ChatGPT offers.
Erik Brynjolfsson, a professor at Stanford Institute for Human-Centered AI (HAI), told Bloomberg earlier in January that “ChatGPT will be the calculator for writing.” Although it won’t replace “thinking and writing,” it’ll enhance those abilities.
Sam Altman himself used the comparison during his conversation with Connie Loizos for Strictly VC: “We adapted to calculators and changed what we tested for in math class … [ChatGPT] is a more extreme version of that, no doubt, but also the benefits of it are more extreme as well.”
I don’t buy the analogy and I think it’s tricky. It implies ChatGPT’s problems are irrelevant. When we accept the comparison, our perception shifts (it’s just another tool in the student’s toolkit) erasing any trace of skepticism. I assume it isn’t intended to be taken 100% seriously, but people will buy it. It’ll spread and once we’ve heard it enough we won’t question it further—it’ll become axiomatic.
It’s attractive because it’s easy to understand, but although there’s some truth to it, there’s nuance, too. To some degree, ChatGPT matches what we intuitively imagine to be “a calculator for writing,” but the overlapping is nowhere near perfect. Metaphors are valuable as long as we don’t stretch them too much (e.g. “the brain is a computer” may lie just in the frontier of the acceptable). Otherwise, they become distracting at best and misleading at worst.
People who propose this analogy use the positive commonalities between these tools to underscore ChatGPT’s usefulness. At the same time, they ignore differences that would put ChatGPT in a not-so-flattering light.
The comparison serves an additional purpose: If ChatGPT is like a calculator then it’d support the idea that, although it feels like cheating now, in reality, it isn’t. Also, it implicitly argues that we’ll come to embrace it eventually—we’ll accept ChatGPT helps (not hinders) students, just like we accept calculators today.
In this article, I’ll explain where this comparison breaks apart and the changes we’d need for it to be true.