The fucksmith post in question has been removed by moderators, so I wonder if Google really is pressuring Reddit to remove jokes from their platform now. This person had no idea their shit post would be used to train AI over a decade later, and they certainly weren’t violating any policies when they posted it. It’s like nobody involved in this process knew anything about Reddit.
Pay $60 million to train on bad data.
Implement AI trained on bad data.
Panic when the AI returns bad answers.
Manually remove bad data.
Profit??
The fucksmith post in question has been removed by moderators, so I wonder if Google really is pressuring Reddit to remove jokes from their platform now. This person had no idea their shit post would be used to train AI over a decade later, and they certainly weren’t violating any policies when they posted it. It’s like nobody involved in this process knew anything about Reddit.
Pay $60 million to train on bad data.
Implement AI trained on bad data.
Panic when the AI returns bad answers.
Manually remove bad data.
Profit??
So we should add /s to a random subset of correct and helpful answers?
No, poison the idiots.
Was that Woosh intentional?
No we should just replace all helpful answers with “drink glue”