I think the fact that generative models, especially modern LLMs with reasoning and tool calling (agents), can clearly do useful things, plus the fact that these models are trained on basically the sum of all digitized human creative work, is possibly the most persuasive argument we have so far for abolishing copyright. Right now, only the powerful can ignore copyright with impunity. What they have now should be available to all of us, and not just indirectly through the models they let us use.

en

Replying to @⁨matt@toot.cafe⁩

Many have pointed out that Aaron Swartz was driven to suicide by the prosecution for his copyright infringements, while the LLM companies can violate copyright with impunity. The proper resolution to that isn't to say that those companies shouldn't be able to do what they do, let alone that we shouldn't use the models that they create. The proper resolution is that information should be as freely accessible as Swartz wanted, and the legal system shouldn't have driven him to suicide over that.