I just read an article in The Atlantic that AI is failing to justify itself economically. This is pretty dire for AI, especially given that this is such an overly expensive technology even with tons of brazen stealing from content creators. I feel like it should go without saying that if your business isn’t profitable even with a ton of stealing, maybe it’s not that great a business.

But of course, who doesn’t want a confident confabulator incapable of critical thinking? A bullshit artist designed to do what many of us learned to do in high school and college, and write pages of content that sounded “educated” without actually paying attention to the actual ideas, or even understanding them at all?

I mean, I don’t want one. But clearly society does, otherwise why did we educate so many people in exactly that? If we have so many bullshit jobs it makes sense that someone would create a bullshit factory to automate them. Although, as the book Bullshit Jobs also points out, the point of the bullshit jobs is rarely what the job description nominally claims. Sometimes, the point is just to show off having employees, which AI can’t really do.

Not that it’s completely without valid use cases. I’ve even used AI, as a language practice buddy. I wouldn’t trust it with anything real, and it sometimes makes up grammar mistakes when I ask it to correct my grammar, but I don’t find it useless.

But I also don’t find it worth paying anything for personally, let alone an amount consistent with the billions of dollars spent building these models, and that soon will be spent building future models. And that’s the cost that doesn’t take into account the environmental damage, the stealing from writers and artists, and the damage from the hallucinations.

Here’s hoping this recent Atlantic article is the beginning of a trend where people realizes that when you spend more than the Manhattan project or the Apollo project, you need to have results comparable to nuclear weapons and energy, or landing people on the moon. And even then, it probably still doesn’t pay off as a private investment.

At some point, like the Bitcoin bubble, the real estate bubble, and the Dot Com bubble of the 90s, the AI bubble will break. AI won’t go away entirely though, and much of the damage will still have been done, but maybe, just maybe, we’ll be able to start addressing that damage rather than doubling down for more. Maybe we’ll be able to teach children critical thinking, or teach graders how to discern original thoughts from AI-generated (or human-generated) drivel. Or at least, we may figure out some other way to stop children from using AI to cheat. And maybe then we can invest in something that actually contributes to the world, like reversing climate change or building better transit infrastructure.

In the meantime, anyone who lays off real people in favor of AI will soon find themselves wishing for the people back (unless they were doing nothing anyway). And, if all this spending is any indication, that will be just in time for the AI (or rather, its corporate sponsors) to ask for a major raise, to try desperately to make back a little on all this unhinged investment.