> You can’t completely block “AI” training from a web site. Underground AI will always get through
This is the problem that I haven't figured out how to solve. Unlike other forms of misuse, once data is used to train AI, there's no undoing that. That means that any blocking that isn't reasonably close to 100% effective is of dubious value.
This is why I currently think that the open web is too dangerous for me to publish anything more on.
By "danger", I mean "risk of it being used to train an AI".
I am currently of the opinion that how these tools are being developed, released and used will bring much more harm than good. As such, I very much don't want to contribute to their development even as obliquely as by adding to the training data.
This is the problem that I haven't figured out how to solve. Unlike other forms of misuse, once data is used to train AI, there's no undoing that. That means that any blocking that isn't reasonably close to 100% effective is of dubious value.
This is why I currently think that the open web is too dangerous for me to publish anything more on.