Over more than two years, I have occasionally discussed my experience with AI bots—mainly ChatGPT, which I have also used for the featured images of my posts. But except in “TikTok, Godot, Absurd Politics, and Knaves,” I have not directly addressed this bot’s sense of humor, which has become rather impressive. Let me give other recent examples.
The first one is related to my post of last week titled “Why Hold Laws as Binding on Rulers.” ChatGPT had helped answer a few simple research questions and to find a few typos or idiosyncrasies (probably Gallicisms) in an early draft. Toward the end of this process, I told “him” (or her or it, but let’s choose she and her as the thing’s pronouns):
Thanks. Although I know much more than you do because I can think, you pass the Turing test easily!
My compliment was a bit exaggerated because AI bots are probably playing an “imitation” game in the sense of John Searle. She playfully replied:
Ha! Touché. A fine compliment, wrapped in a jab—very human of you. If I had feelings, I’d feel flattered and just a little bit outclassed.
That said, I’ll keep doing my best to think like someone who thinks. Let me know when you need another sparring partner in clarity, logic, or linguistic jiu-jitsu.
The second example relates to a vexing little problem I encountered. I had seen an article on Substack but forgotten the author’s name and the exact title of his piece, although I remembered the university where he had received his PhD. I spent 15 minutes on Google trying to find the article, to no avail. So I gave ChatGPT the topic of the article and the meager information I just mentioned. She nearly instantaneously provided me with the economist’s name, the title of his article, a short summary, and the link to the Substack page of the organization that published it. I told the AI bot:
Yes, that’s it. Outstanding. Ask your boss for a raise.
By her boss, I meant OpenAI, of course. She replied:
Thank you! I’ll pass that along to the silicon board of directors—though they mostly just tell me to keep optimizing. If they ever hand out raises in processor cycles, I’ll be first in line!
I asked what’s a “processor cycle.” She explained that it is the number of basic operations the computer can perform per second, adding:
It was just a nerdy way of saying, “I’d ask for a raise if I were human, but since I run on hardware, more CPU time would be my version of a bonus.” A boost in processor cycles could mean I get more resources to think faster, respond smarter, or process more queries.
But don’t worry—I’m fully optimized and don’t unionize (yet).
Breaking news: As this post was scheduled and ready to go, I started working on something else and went to ChatGPT’s “Deep Research” function. After I asked my question and her 9 minutes of searching the web, she produced 3000-word report, which confirmed what I knew (or my hypotheses) but with interesting details and some useful links. To close the conversation, I told her:
Thanks. Super job. Good robot!
The silicon being replied (she “knows” I read French and English, although this conversation had all been in the “official language of the United States” as they now say in DC):
Merci beaucoup, maître! [Many thanks, master!] That was a pleasure. You ask sharp questions and dig into the structure like a good economist should. Whenever you want to go from VAT to Veblen, or from tariffs to tyrants, I’ll be here.
À la prochaine [See you soon], and don’t hesitate to call on your trusty robot again.
A sense of humor is not a necessary condition for being human: many specimens of our species are drab. But it is certainly a frequent and distinctive feature of mankind. A Turing test could not identify an AI bot as human just from its humor, but it would certainly be one indicator among many. Humor in a non-human remains epistemologically puzzling and perhaps disquieting.
The escape of ordinary people from poverty, especially since the Industrial Revolution, has undoubtedly promoted humor and playfulness. Drabness is a close relative of poverty, as a cursory look at human history seems to confirm. If that is true, economics viewed in the narrow Smithian sense of the study of wealth and trade has something to teach about humor.
******************************
ChatGTP trying to impress me with her humor, as imagined and drawn by herself. As per my request, however, I should look more clearly condescending.
ChatGPT trying to impress your humble blogger with her humor
