Copyright (or IP law more broadly) does a variety of things, good and bad. I’d like to see us keep the good and ditch the bad.

The good:

1. Accurate attribution. It’s important to know who actually came up with something, so you can continue the conversation with them.
2. Faithful replication. It’s important to know what the canonical version of a thing looks like, so you can identify what has changed about any derivatives.

Until AI, these were pretty trivial concerns.

You could easily Google something to find out who made it. And computers produce exact copies by default, so how likely are you to run into a malformed approximation of the original thing?

But now, these are both under serious threat.

The bad:

1. Keeping culture and knowledge away from people who could do something with it, until they pay up.
2. Preventing interoperability.
3. Coercing authors into signing over the rights to their concepts, so they can’t continue working on them unless it’s profitable for the rights-holders.

You’ll notice I left out the most obvious thing: “It gives creators a legal monopoly to prevent them getting out-competed by wealthy vulture corporations copying them.”

That’s cuz as long as they can pre-emptively force creators to sign over the rights, copyright doesn’t protect creators — it does the opposite.

I think AI only gets rid of the good parts of copyright, while doing nearly nothing about the bad parts.

We desperately need copyright reform. But it should not be rewritten by big AI interests.