A black man, Maurice Malone, asks if ChatGPT is racist after receiving a slang-filled essay
Artwork by Saifullah Ali

Is ChatGPT racist?

Is ChatGPT racist? Let me share one of my early encounters with AI software. I was sort of giving ChatGPT somewhat of a test drive, when I asked it to write a beer review of a particular brand in the style of Maurice Malone for my blog

I’ve been writing about denim for a long time and must have written hundreds of articles. I figured it would crawl my articles and pick up some tips on my writing style. I also provided the URL of this blog so that it could crawl my beer-related post.

Was I surprised when the first line of the review began with “YO, YO, YO” and ended with “Peace out?” I laughed while thinking, “WTF?” I don’t use slang like that. This does not sound like me.

I showed it to my brother-in-law, who laughed as well, before quickly surmising that the AI had picked up on the fact that I was well-known in hip-hop culture and went from there.

I sell denim and make fashion, but one of my most important jobs is writing content because I also build and maintain our websites. I do so many things in a day that my time is extremely limited, so I’ve become very acquainted with using AI writing programs to assist me.

The keyword for me is “assist,” not “write for me.” The first priority on my list is, of course, spell checking, followed by better ways to phrase something or better word use, and so on. I frequently find that AI can run on and on without saying anything meaningful, spitting out bland paragraphs of whatever they scrounged off the internet, and sometimes not factual. Most importantly, I find AI useful for inspiration (finding a new angle) and getting to the point of my thoughts better.

Back to my ChatGPT test drive, where I gave it a thumbs down and wrote a few paragraphs in the comments section. Essentially, I expressed my displeasure that the software chose to generalize my writing style rather than analyze the specific articles I had written.

I repeated the test about a month later and discovered that either ChatGPT was learning or my comments had some real impact. I’m happy to say that I did not get the same results. The program started by outlining its limitations. After my inquiry, there was no slang in the new results of a similar request, and I got real, useful content to help me.