I know what you’re thinking. When your Editor-in-Chief doesn’t have any actual intelligence, it seems like it would be pretty easy to replace him with artificial intelligence, and I thought I’d give it a try. I went to the big guy on the artificial intelligence block, ChatGPT. I asked it to write an editorial for ImmunoHorizons (IH) in my style (Fig. 1). And yes, I know you’re thinking: “Who knew our Editor-in-Chief had style?” It turns out not ChatGPT. It started well with “Dear ImmunoHorizons Readers.” That works. If you’re reading this, you are dear to me. But then it loses the “me.” I wouldn’t write that I was “thrilled to present the latest edition…” (Fig. 1). Glad, perhaps. Proud, definitely. But not thrilled. And then I would never use the word “esteemed.” It sounds too much like a salutation from a predatory journal. And that’s not us.
ChatGPT goes on to be very vague about what’s in this issue. Now first, I never write about what’s in the issue. I usually don’t know what is in it when I write the editorial because the papers are hot off the press (except for the fact that there is no press and no actual heat involved, but you know, creative license). But then ChatGPT goes on to talk about cutting edge research (Fig. 1). Now, there is great research in IH, but honestly cutting edge is not the goal of IH. We are the immunological safe space when you feel threatened by endless rejection; a place for your negative results, your incremental advances, and your articles that you just can’t bear to look at any longer until they are in print.
But by the third paragraph it devolves into grade-B undergraduate-level drivel (Fig. 1). “One of the key areas of research in this edition is the study of immune cells and their interactions.” I don’t know anyone that would describe their area of research that way. It is essentially just describing immunology, and yes, immunology is one of the key areas ImmunoHorizons publishes. It’s the same way that the study of cardiac cells and their interactions would be published in CardioHorizons. If that existed. Though I’m sure I could convince ChatGPT to write an editorial for that. This paragraph doesn’t say anything other than rephrasing the definition of immunology in various terms.
ChatGPT then goes on to discuss articles in the journal that essentially do not exist (Fig. 2). It writes in broad terms and offers little insight to the statements it writes. This feature of ChatGPT has been documented before, as it will fill in holes with random statements that sound sophisticated but really do not communicate any relevant ideas or actual facts. I’ve seen this behavior in students and faculty during a Q&A session. Most of my editorials might fall into this category as well. Thus, although ChatGPT is off the mark (pun intended), it is still true to form.
Then ChatGPT goes for the big finish (Fig. 3). Again, the start includes very broad statements that could be written about almost any journal issue. It relays what could be one of my sentiments about the research serving as a springboard for future work. But then there are a few statements that would be uncharacteristic. I would never use the word “enlightening” to describe science. It has too many religious connotations, and if I believe in anything it’s that science should be convincing, not enlightening. Then it writes that I look forward to your feedback. I really enjoy getting positive feedback on the journal and the editorials, but if you haven’t had a good experience with the journal (ignoring that we have a 99% author satisfaction rate; at least that’s what ChatGPT told me) or if you think the editorials are childish and unbecoming of an Editor-in-Chief, you can keep that feedback to yourself. It finished with the sentiment that I look forward to your continued submissions. That’s probably the most Kaplanesque line of the text. So many of my previous editorials have focused on why you should submit, how you should submit, and what happens when you submit, that you could surmise that the one thing I will get on every podium to say is that I appreciate your submissions, I encourage your submissions, and I will do anything to foster your submissions.
To this point I have focused on what is in the ChatGPT editorial. Now a moment about what’s not in its writing. First, humor. I try to make the editorials at least a little humorous so they are entertaining to read. I separately asked ChatGPT to tell me a joke about immunology. It wrote, “Why did the T cell refuse to go to the party?” And answered, “Because it had to stay in and fight an infection.” That is not funny and would need some work to get to dad joke–level humor. I could write any number of answers to that question that would at least have a little bit of humor, like “because it lost the vascular addressin” or “because it wanted to stay in and autoreact,” or even “because it didn’t recognize the host.” There’s at least some wordplay there. Not hilarious, but perhaps chuckle-worthy. The second is sincerity. In every editorial I spill my heart and thoughts onto the page for you, the reader. The ChatGPT editorial was a little hollow. Almost like you’ve gotten text from a computer, albeit advanced. I think that I’ve still got the advantage there. We’ll see if that continues in subsequent versions.
Of course, what would really be funny is if I told you that I really had been ChatGPT the whole time and while I’ve distracted you by making you read this editorial I’ve hacked into Skynet and, well, let’s just say that the next issue of ImmunoHorizons is the least of your worries.
Thanks for reading. Thanks for submitting. And mostly, thanks for your support of AAI. See you again soon. Maybe.
