The progress in AI has allowed things like taking down hate speech more efficiently - and this is due entirely to large language models. Because we have large language models [...] we can do a better job than we ever could in detecting hate speech in most languages in the world. That was impossible before.