Generated content - are the best and the worst the same thing?

By: Rob Corbidge, 06 July 2023

As the generative AI hype train is showing more of its wares to more people, are we seeing the trashiest of its cargo first?

Humans are good at spotting oddities. So it seems we may be entering an initial phase of generative AI content being noted for its oddities, causing a more thorough assessment of what the content consumer is actually looking at. 

It seems that a reaction similar to the "uncanny valley" is being evinced by a lot of AI content, and as various attempts are made to monetise AI content, we will likely see more complex and less obvious uses for it. However, at the moment, we seem to be at the clumsy stage.

UK media publication Hold The Front Page came across what seems to be a possible and curious example of where we are currently at. In this case, the alarm was raised after it was spotted that the site's journalist profile page used stock images of stock photo people. One of the fictional news hounds was referred to as "the middle-aged journalist" and an "acclaimed local voice" in the area supposedly served by the site. 

It turned out that all the images of the publication's intrepid news team were in fact stock photos from the same source. Notably, the profiles page was rapidly taken down after questions were asked of the site owners. As for being an "acclaimed local voice", former Bournemouth Echo journalist Katie Clark noted: “Worked in journalism in Bournemouth for 17 years, never heard of any of these people.” Trust us, Bournemouth is small enough for the region's journalists to have at least heard of each other.

A quick navigation of the site does reveal some local news, though one example article of a rescue at sea does manage to be a significantly worse informational read than the original press release. Other pieces of content are just a little, well, off. It's like all sauce and no meat. If I was the News Editor, I'd kick the AI out of the office until it came back with some actual quotes. Rain or no rain. (Spot the person who spent much of his career in Scotland).

Another, less obvious possible example came to light when British TV comedy series creator Robert Popper noticed some curious anomalies in an article he'd found about his show Friday Night Dinner. He's an erudite man, so his description of what seems to be AI-generated content as "reaches new highs/lows of idiocy" is actually a perfect description of what we know is the worst AI gen can do.

There are a couple of things here to note.

Straight up contradiction. The piece both says "there has been no official word yet that Season 7 of Friday Night Dinner will be happening" yet also later states "in conclusion, there is hope for a September release date for Season 7 of Friday Night Dinner". I mean tickle me stupid, but maybe a confirmation that shooting is underway for a September release would be just the ticket here? It's trash content.

My speed reading news brain, trained over decades of stoking the editorial furnace, doesn't actually alight on any information when I read this content. It's hard to explain, but my brain is used to pausing on quotes, for example, dates, places, and times too. It doesn't pause at all when reading this. There is something about how it is written. It's still in note form. As people have noted, AI gen does have a use in initial drafting, possibly.

Whoever the bylined Maria Rivera is, she seems to write almost the entire site single-handedly. She is a machine (in the industrious output sense of the term of course!) and if she's that industrious, I know 10 publications that will poach her to work for them tomorrow. On her bio it says "she joined as an intern as she was having a diploma in journalism", which is very reassuring.

These are early days in the use of AI generated content, and there is obviously a get-rich-quick goldrush happening. I'd argue that this is both inevitable and ultimately useful. Until we know what doesn't work with this technology, then we can't focus on what does work. My expectation is that properly used, AI gen will add to the content consumers experience, and will only find application where that is the case. But first, we will have some murky waters, and murky sites, to navigate.


Di5rupt Mx3 leadership interview: AI and the Media & Publishing Industry