Is Everything Fake?
Are you worried that ChatGPT is about to become sentient and turn you into a paperclip? Probably not.
What worries me is that our culture is in danger of becoming overrun with spam.
Because ultimately, that is the main use case for LLMs: zero-effort writing. Need to write a report? Chat GPT will spit one out for you.
Universities are facing a crisis because many of their students can't be bothered to write their own essays and are farming the work out to LLMs. Websites are springing up everywhere, filled with AI-generated slop on every topic imaginable in the hope of chasing advertising money.
Workers in offices are getting ChatGPT to write reports for them, and their colleagues are getting ChatGPT to summarise them.
Google results can no longer be trusted. You follow a link, only to start reading and notice that something is off: the information is vague and the writing waffles aimlessly, often repeating itself. It's AI slop.
And it will only get worse as the LLMs start feeding on each other. An Ouroboros of bullshit, gradually getting worse with each generation.
You can't trust anything you read, anything you see, or anything you hear.
You can't even trust that this newsletter you are currently reading right now was written by a human.
Many of the newsletters that are now landing in your inbox are machine-generated.
This one isn't. Aside from Grammarly (whose suggestions I ignore at least half of the time), I don't use any form of AI in my writing. Why? Because an LLM knows nothing about me, or my life, or indeed about you either.
The bots can write generic content that might work as filler, but nothing that can really create a genuine human connection.
We don't need more "content". There is enough marketing slop, enough unread novels, enough overlooked art and music in our culture already. We don't need more of this stuff.
Machine learning could be incredibly useful. Why not train it on some of the tedious stuff? To retopologise a sculpt. To create perfect UVs. You know, the stuff that artists find cumbersome and would love to have some help with. There are endless positive potential uses for this technology.
But don't make art for me. Don't write for me. That's the stuff I want to do, the stuff that is meaningless without the human touch.
But most of all, don't destroy our trust.
Don't make us suspicious.
I need to know that a human being cared enough about their art, about their writing, that they took the time and trouble to do it themselves. Because if they didn't, if they cut every corner they could, why should I engage with it?
So my fear about AI isn't that we are about to become slaves to some hyper-intelligence, or that it is coming for our jobs.
My fear is that it is being used to create a firehose of bullshit that is eroding trust and quality on the internet. And now that the tap is on, I'm not sure how we can turn it off.