A key challenge for purveyors of AI text generators is finding large bodies of text to train them on that aren't full of garbage. But the makers of GPT-4chan just went right ahead and created one fine-tuned with 4chan.
"This thing is vile," warns the prompt form, should you wish to experiment.
There is a point: zero-shot toxicity detection. But what a fine moment to remember that when Nietzsche wrote about looking too long into the abyss he was writing about the danger in fighting monsters.
GPT-4chan is trained on anonymously posted and sparsely moderated discussions of political topics. Its intended use is to reproduce text according to the distribution of its input data. It may also be a useful tool to investigate discourse in such anonymous online communities. Lastly, it has potential applications in tasks suche as toxicity detection, as initial experiments show promising zero-shot results when comparing a string's likelihood under GPT-4chan to its likelihood under GPT-J 6B.