* Posts by mdfischer

2 publicly visible posts • joined 12 Jan 2024

OpenAI: 'Impossible to train today’s leading AI models without using copyrighted materials'

mdfischer

Re: Sounds like...

The publishers do put pretty much all their books online. If OpenAI didn’t have the good sense to pay for these, then they deserve whatever problems they incur. I’m sure that the NYTimes did sell a subscription to OpenAI, and probably a commercial one like they would sell to a University. Will the times go after the universities to collect their cut of the lifetime efforts of their students?

mdfischer

Copying content is a problem. Copying style is not so long as you don’t try to pass yourself off as the person whose style you cover, which is a criminal offence, not civil like copyright. Every writer, painter, musician, and ordinary person learns from the work of others. I don’t think we would want to monetise that.