General Discussion

I recall.

I usually enjoy Cory's social media posts and newsletters, but don't agree with him on everything, including his views on copyright.

And IMO the first problem with genAI is the theft of copyrighted work for training data. That law-breaking and exploitation of workers defined the AI bros' approach, and the only really good solution is destruction of all current illegally trained AI models, training of new models with only what's in the public domain and what the AI companies have legally acquired from the copyright owner, and punishment of the IP thieves including prison sentences.

Cory just mentions sectoral bargaining, which he admits is currently illegal for most workers.

Most of the value of genAI tools derives from the IP theft. Destroy illegally trained AI models and compel legal training of new models, and suddenly genAI has very little of value to offer, unless AI companies are suddenly willing and able to pay for much of what they stole earlier. It will destroy any business model for genAI, and funding for it.

Especially because LLMs are inherently flawed, their results untrustworthy.

And in the.meantime, opponents of illegally trained genAI should do everything in their power to make its use socially unacceptable. Not just the initial theft, but the foolishness and callousness of users who don't care about that theft, should be exposed and condemned. It's no more a legitimate, ethical business than the slave trade was.