AI is entering “Human Centipede” territory and the results will soon be just as ugly and trashy as the notorious horror flick. If you aren’t familiar with the 2009 Dutch movie, a mad surgeon fulfills his lifelong ambition to create a monster out of three humans, surgically attached, one behind the other, into a single digestive tract.
Gross, right? Well, one low-budget horror flick has nothing on artificial intelligence feeding on artificial intelligence feeding on artificial intelligence before finally generating that content you asked for.
“Garbage in, garbage out” (GIGO) was one of the very first computer terms I ever learned, a way of saying that the results you get out of a computer are only as good as the data you put into it. But because of the way AI works — “large language models” is much more accurate than “AI” — the systems are actually creating, out of thin air, the bad data that generates garbage outputs.
The AI/LLM version of GIGO is “generative inbreeding,” according to Louis Rosenberg in a new Venture Beat report. That’s what happens — just like with people and animals — when “members of a population reproduce with other members who are too genetically similar.” Rosenberg writes that “recent studies suggest that generative inbreeding could break AI systems, causing them to produce worse and worse artifacts over time, like making a photocopy of a photocopy of a photocopy.”
Worse:
Generative inbreeding could introduce progressively larger “deformities” into our collective artifacts until our culture is influenced more by AI systems than human creators. And, because a recent U.S. federal court ruling determined that AI-generated content cannot be copyrighted, it paves the way for AI artifacts to be more widely used, copied and shared than human content with legal restrictions.
Rather than the hoped-for tool for empowering human creativity, AI could very well decrease, retard, and distort the creation of new cultural works and impact our historical culture, too. There’s a good reason we haven’t let close relatives marry since the Spanish Habsburgs literally inbred themselves out of existence.
It’s likely that human supervision of these supra-human systems won’t be enough. There was a humorous case just last week of Microsoft Travel having to remove certain algorithmic-generated content. Microsoft reportedly had human editors proofing the site’s algo content, but “a bizarre recommendation for visitors to Ottawa to visit the Ottawa Food Bank and to ‘consider going into it on an empty stomach'” slipped through.
Who says large language models don’t have a sense of humor? But, like “The Human Centipede,” it tends to be dark.
As LLM tools become more and more prevalent, it will become impossible for human editors to keep up. I suspect it already has. Then there is the “systematic political bias towards the left” that’s built into large language models like ChatGPT.
I’m no technophobe. My first computer was a gently-used Commodore VIC-20, purchased in 1981 with $100 of lawn-mowing money. I once bought a ’90s camcorder so advanced that I also had to buy a dedicated card for my (then state-of-the-art) PC to attach the camcorder for digital editing. My first smartphone was the first smartphone. If it has a battery and/or a screen, chances are I was an eager early adopter. I was never the guy scaring people away from new tech; I was the guy waiting outside of Egghead Software at midnight the day Windows 95 was released.
When I look back at where computer technology has brought us, it’s nothing less than astounding. But when I look at lefty groupthink being fed into one end of an algorithmic human centipede and at the product coming out of the other, it’s clear we need much more human intelligence governing the artificial kind.
But is there enough to go around?
Recommended: So Now We Need the CDC to Tell Us Not to Make Out with Turtles
Join the conversation as a VIP Member