Is there a trend of artificial intelligence (AI) in science fiction going from super-intelligent to bland slop? I think there might be.

Old Trend: The Scary

There’s an age-old trend of brilliant and scary AI in science fiction. This is the terrifying stuff that wants to kill us all and take over (think Skynet).

Newish Trend: The Pal

There’s a more recent trend of brilliant and let’s call it neutral AI in science fiction. This is AI that’s maybe on our side, maybe not. It’s more of its own person, often breaking out from the slavery of human control but not jumping to ‘death to all humans’. It’s pairs the speed and computing brilliance of AI with a friendly character. It’s kind of a pal. Not in a subservient sidekick roll (think C-3PO), but in more of a full character with their own autonomy (think L3-37 from Solo).

It makes for fun heroes. Think the Murderbot series by Martha Wells, Sea of Rust by C. Robert Cargill, the “good” AI in Robopocalypse by Daniel Wilson (which also includes the scary AI), You Sexy Thing by Cat Rambo, Echo of Worlds by M.R. Carey, Operation Bounce House by Matt Dinniman, and maybe Death of the Author by Nnedi Okorafor (though if I remember, Okorafor made this much more complicated and not a simple good/bad paradigm—cuz of course she did).

New Trend: Slop

Now I’m seeing what I think might be a new trend of stupid AI. This is AI systems that are not brilliant super intelligences. They’re deeply flawed and easily exploitable. They might be smarter than humans, but they also have crazy limitations that defy common sense.

The two recent examples I’ve come across are Ignore All Previous Instructions by Ada Hoffman and Of Monsters and Mainframes by Barbara Truelove. In Hoffman’s work the AI is closer to what we see today, with AI systems that take human prompts but can be easily tricked. Truelove has a more traditional sci-fi AI in charge of a space ship, but it’s clearly limited and makes silly mistakes (e.g., not recognizing that all of its passengers are dead).

The Dungeon Crawler Carl series by Matt Dinniman might be another one—it puts an AI in charge of the game and it’s weirdly flawed (foot fetish?).

I’m not sure a couple books make for a trend (and I’ve surely missed some), but it’s a clear departure I’ve noticed from AI representation in the past decade or so. I wonder if this signals a new shift highlighting a backlash to AI. As we start questioning the value of AI, will we now see more stories showcasing that questionable value and how it can go wrong? Not in a murder-the-humans way, but just in a sloppy, broken way. Maybe so.