As a programmer, I am exposed to other professionals who just love AI. They regularly employ the most tired and worn technology clichés like “game changers” and “transformative” like they’re not tired and worn. As with plastics and PFAS, the topic leeches into everything tech—and sometimes literally everything. For instance, there are people out there bonkers enough to feed their medical data into ChatGPT and, as it turns out, it’s as unreliable as it ever was1.

A pattern I continue to notice from the LLM camp when criticism about the accuracy or utility of the responses is raised always ends up being some variation of the classic “you’re doing it wrong” argument. Maybe your prompts aren’t clear enough. Maybe you need to add another 10,000 lines to a context file. Maybe you are expecting too much. Whatever the case, the problem is clearly you for misunderstanding that a technology being lauded as “transformative,” being called “intelligence,” and peppered up and down with enough anthropomorphism to make you unable to believe it’s anything but a some kind of digital buddy or assistant despite the fact that it is absolutely none of those things.

What? You thought you were going to interact with a large language model using plain language and get meaningful, consistent, results as one would expect from an entity possessing so-called intelligence? Well, joke’s on you. What are you? A fucking rube?

For a moment, let’s forget all the cultural, environmental, informational, social, corporate, and economic rot caused by this technology—and no, you don’t get to absolve yourself of your part in all that if you use it anymore than any other petty evil in your life. While we’re tossing realities out the window, let’s also remember that no one using this stuff is currently paying anything close to rack rates because it’s all subsidized so… I guess we need to forget about that too. Right, so let’s just think of it as a tool sans the baggage.

It’s still bad in terms of what it proclaims to be and it’s bad in terms of its methods. If I had to fight with my chosen programming language, text editor, or even something as mundane as a spreadsheet application to try and convince it to do its advertised job… I would hate it. (I specifically stopped using Microsoft Windows on this very basis like a quarter of a century ago.) No one but the most delusional would be making excuses for it. Now add back in all the abysmal realities and I have to ask what is becoming my most oft asked question…

What the fuck are we even doing?

  1. The Washington Post, reporters, and most of our healthcare professionals are probably also very unreliable because pretty much everything is overworked, corrupt, lazy, or sensationalized (or owned by Jeff Bezos). I don’t want anyone to think I put doctors up on any sort of pedestal—unless I mean to have crowds pelt them with rotten fruit. Healthcare sucks and the arrogance wrapped in lab coats is a huge factor. ↩