Hmm, how can one write intelligibly on the (web) walls for all to see AND at the same time prevent the 'datavores' (@thegibson) from devouring it and getting stronger by it???
(referring to chatGPT and all..)
@hanspeter A question that often comes up and the answer is: it's difficult. When you put something out for everyone, the datavores will not shy back from using your content for their products. Putting something on a big platform won't help either because they'll sell your content to the datavores. I can only think of two options:
Use the Smallweb (gopher, gemini, spartan). But I assume it's only a matter of time until the datavores will use that too
Put your content behind a login and send out invites
I often hear that a non-AI-license will do it, but I doubt that any robot really cares. They'll suck in your content and simply don't care about the license at all and you'll never be able to prove that they violated your license.
So in the end, IMO if you want your content to be seen by all, that always includes the robots and there is nothing you can do about it until there is proper legislation.
@louis@hanspeter@thegibson I have been meaning to create a procedural nonsense generator from something like GPT2 or some LLM that is small enough to run on a laptop. Then I am going to flood my online content with nonsense at a 1/100 signal/noise ratio and let the LLM scrapers have at it. Of course I'll make it easy for a human to filter out, by placing the noise content in hidden <div>s, or by removing it with a bit of JavaScript, or coloring it the same as the background.
Gopher is a good idea too, I could put non-obfuscated versions of my documents there.
Add comment