Ulisse Mini comments on How I failed to form views on AI safety