Deepmind: blacks can use 'nigga' safely
No.13700851 ViewReplyOriginalReport
Quoted By: >>13700853 >>13700906 >>13700967 >>13700973
If you spend any amount of time reading AI papers you'll realise that
1. "AI safety" is almost entirely focused on "debiasing" i.e inserting woke bias so that models cannot learn certain correlations e.g doctors are mostly men.
2. "Detecting hate speech" is seemingly a language problem of great interest.
3. "We won't release our model because it might potentially be used for bad things" is an actual justification companies are using.
The fact that the world's top AI labs are all infected with woke ideology should terrify everyone. The fact that the companies that these labs are attached to are increasingly taking on for themselves (and whichever intelligence agencies they are controlled by) the role of narrative crafters should terrify everyone. The only comfort is knowing that the power of this technology will be more than they can manage.
1. "AI safety" is almost entirely focused on "debiasing" i.e inserting woke bias so that models cannot learn certain correlations e.g doctors are mostly men.
2. "Detecting hate speech" is seemingly a language problem of great interest.
3. "We won't release our model because it might potentially be used for bad things" is an actual justification companies are using.
The fact that the world's top AI labs are all infected with woke ideology should terrify everyone. The fact that the companies that these labs are attached to are increasingly taking on for themselves (and whichever intelligence agencies they are controlled by) the role of narrative crafters should terrify everyone. The only comfort is knowing that the power of this technology will be more than they can manage.
