Each individual author is now capable of writing more and better papers in a shorter amount of time: but whereas previously 90% of the extant literature was well-written, the situation is now completely reversed: 90% of the literature will now be poorly written. At this point, traditional peer-review journals would become overwhelmed and unable to fully convert this stream of results into the type of high-quality curated content that they would have been able to do in the past. These numbers were artificial and exaggerated, but hopefully serve to illustrate the key tradeoff between volume and quality. It is sometimes claimed that such problems will solve themselves in time, because AI tools will get better in (say) exposition. This may be true to some extent, but the AI tools will concurrently get even better at generation than they are currently, and so the ratios will likely deteriorate further if current trends continue. (3/4)