A few weeks ago I found a post on another site where someone had asked an AI to write a film review. The results were what I think you might expect: a bland, clichéd summary of opinion such as you’d get from a review aggregator.

The reason this is what you’d expect is because the way these programs work (and I’m aware that people who understand this field better don’t even consider it to be AI) is to just take all the data there is on a subject and melt it down to something that sounds like a general consensus. So of course it’s going to be clichéd and derivative. Cliché is, by definition, the most common form of expression in the datasets from which it draws on.

What we’re left with is the hive mind, which is where we were heading anyway what with review aggregators and the like. The “wisdom of crowds” is a distillation not of the best that has been thought and said but of everything that’s been thought and said. And I think for a lot of people, and for different purposes, that may be good enough. For people who read genre fiction by the bale, those looking for executive summaries of generally held views, or students looking for a precis.

In the field of aesthetic response or opinion writing, however, is this the best we can expect? I started thinking about this because of an article I read online at the Yahoo! Sports page covering NFL football. I originally pulled a blank on the byline “Castmagic.” Was that a person? People have lots of strange names these days so I thought it possible. But when I clicked on the link to read it I found this:

(This article was written with the assistance of AI and reviewed by our editorial team to ensure accuracy. Please reach out to us if you notice any mistakes.)

I immediately had some questions. It was just a short opinion piece, so what did it mean that it was written “with the assistance of AI and reviewed by our editorial team”? My own sense was that it was written entirely by AI and just proofread and copyedited for factual errors or anything that might get Yahoo! in trouble. I also wondered if this was a direction more news organizations, and not just Internet ones, were going to be heading in.

The subject of the piece was the New York Giants football team. The Giants were a very bad team last year, resulting in their having a high pick in the upcoming draft. They don’t have a clear starting quarterback on their roster and it’s usually assumed that a team in such a situation will pick the best QB on their draft board as this is the most important position to have filled. So the question posed to “Castmagic” was “Is it time for the Giants to draft a quarterback?”

Things didn’t get off to a good start: “As the dust settles from the 2024 NFL season, it’s evident that some teams face more pivotal offseasons than others.”

Well, duh. We’re hit in the face with a cliché right off the bat, followed up by an obvious truism. I didn’t need an AI to tell me this.

As “Castmagic” went along it mostly borrowed from an earlier column on the same subject written by one of Yahoo!’s (human) sports writers. But if I’d been that particular writer I don’t think I’d look at this as being the sincerest form of flattery. I’d probably be worried for my job.

Did “Chatmagic” have any original insights to offer on the question of whether the Giants should draft a QB? No. Here’s the conclusion:

In the end, the Giants’ path forward hinges on navigating the delicate balance between short-term success and long-term strategic planning. Whether through drafting a quarterback or trading down to solidify the entire roster, the Giants face decisions that could define the franchise for years to come. Only time will reveal if they choose wisely.

Really? That’s the takeaway? The Giants have options and “only time will tell” if they make the right choice?

Will “Chatmagic” get better? I think it will, if only because I don’t see how it can get any worse. Or less useful. But I think these early, baby steps give some indication of the issues going forward, at least when it comes to this form of writing. How can an opinion of any value on any subject be fashioned out of a dataset that is just a collection of everybody else’s opinion? These programs aren’t interested in original insights or finding out the truth. Are they even capable of that? Only time will tell . In the meanwhile, what we have now reads like a page of Google search results, just the repackaging of random information, some of which is no doubt total garbage, into a paste of content that you can skim your eyes over before clicking onto what’s next.

There were some 50 comments on the article the last time I checked. Most of them piling on the “dummies” who write sports opinions for Yahoo! Only one of them registered that it had been written by an AI.