Somebody shared an image with me a while back that I can't get out of my head. They were talking about a tilapia farm.
If you're not familiar with how a tilapia farm works, here's the gist.
Thousands of fish are packed into a pool so tight they can barely move. You drop food in. They eat the food, and they poop. Then they eat the poop. Then they poop out the poop they just ate, and they eat that too.
When one of them dies, the rest eat their friend. Then they poop their friend out. And the next generation eats a pooped-out version of their old friend.
The whole farm is a closed loop. Same fish, same pool, same food cycling back through them in degraded form, generation after generation.
That's a tilapia farm.
A closed loop. Same fish, same pool, same food cycling back through them in degraded form, generation after generation.
Recently when I was at Strategic Coach, we were talking about artificial intelligence. Another entrepreneur shared their fear that AI is becoming like a tilapia farm.
I was already familiar with the concept. Years ago, when a friend saw me ordering tilapia, he told me how it's farmed. I haven't ordered one since.
This framing for the dark side of artificial intelligence is hard to shake out of your head once you see it.
This framing for the dark side of AI is hard to shake out of your head once you see it.
Full disclosure: I am a huge fan of AI. I genuinely believe it's one of those tools, like many before it, that will help humanity grow, prosper, and improve quality of life. But just like anything else, no matter the benefits, every tool has drawbacks and requires guardrails. Smartphones have done incredible things for us, and they've also fueled addiction and a bunch of other setbacks we're still working through. AI is no different.
The reason the tilapia farm stuck with me is because we've all already experienced AI like this. My favorite example is the person who told me they automated all their email. Every single email gets replied to. Well, if you automate all your emails and I automate all my emails, then it's just artificial intelligence talking to artificial intelligence. We're all eating our own poop.
We're all eating our own poop.
If you want a less disgusting analogy than a tilapia farm, this all reminds me of a cinematic classic: Multiplicity. (Starring Michael Keaton, or as we say here in Pittsburgh, Batman. And let's settle this once and for all. There is only one Batman movie that matters, and it's the 1989 Batman with Michael Keaton.)
In Multiplicity, Keaton's character figures out how to clone himself so he can better juggle all 168 hours of his week. He sees how a second version of himself could ease the stress, take some of the load with the wife and kids, and let him get more done. So he does it. The first clone is great. Then the clone realizes the same thing and makes a clone of himself. That clone makes another. By the fourth generation, "a copy of a copy of a copy," each version has fewer of the original's great qualities. They're a mess. Less sharp, less capable, less like the man who started it all.
That's the warning for how we use AI.
By the fourth generation, each clone is a mess. Less sharp, less capable, less like the man who started it all.
Large language models are trained on human work. As more of us use a small handful of AI tools to produce a flood of content, that content goes back online and becomes the training data for the next version of the model. AI ends up learning from AI, which learned from AI, which learned from AI. A copy of a copy of a copy. A closed loop feeding on itself.
Sound familiar?
A copy of a copy of a copy. A closed loop feeding on itself.
Some of that evolution is natural and even good. Human culture changes. Language shifts as people move and mix and remix what they know. AI is going to be part of that. And if you know how to prompt well, how to guide and lead AI as a thought partner, you'll get sharper, more original results than someone who doesn't. We're not all going to end up with the same output.
Then comes AI slop.
You've seen it. Someone types a prompt, copies the result, and ships it. No checking, no refining, no shaping it toward what they actually had in mind. They let AI be in control instead of the other way around. As more people do that, more slop fills the internet. That slop becomes the next round of training data. And the quality of every model that learns from it gets weaker.
The slop becomes the next round of training data. And the quality of every model that learns from it gets weaker.
So let me ask the question. Are you doing the same thing with AI? Are you feeding the loop?
Are you feeding the loop?
This is a real possibility, and it's already happening. But it's not something to be afraid of. It's something to be intentional about.
The answer isn't to fear the tool. It's to lead it.
The answer isn't to fear the tool. It's to lead it.
The point of AI is not more output. The point is higher quality output. The point is not serving more clients. The point is serving your clients better. AI should help you take what God has placed on your heart, the knowledge in your head, and the vision you carry for a brighter future, and bring it to reality with greater impact than you could on your own.
Don't use AI to solve more problems. Use it to solve bigger problems.
Don't use AI to solve more problems. Use it to solve bigger problems.
That's how we turn this from a tilapia farm into something more like a wild salmon run. Where what you put out into the world is stronger, wilder, and more nourishing than what came before. Where your work compounds in value instead of decaying in copies.
If you're going to use AI, partner with it. Lead it. Refine it. Make sure what comes out the other side actually reflects you, your standards, and your calling. Otherwise, you're not raising salmon. You're raising tilapia.
Otherwise, you're not raising salmon. You're raising tilapia.
Live Intentional.