... so? That happens to every new technology. It is mostly useless until some development makes it viable or useful. Getting a time cut from 2019 to 2023 and

... so? That happens to every new technology. It is mostly useless until some development makes it viable or useful. Getting a time cut from 2019 to 2023 and extrapolating that now it will advance at an exponential pace every few years is as stupid as getting a time cut from 1980 to 2020 and predicting it will have a glacial pace of advancement in the future. Truth is that there were some interesting advancements in LLM research, but it's difficult to tell what will happen. Will it plateau for a while? Will it keep advancing at a fast pace? Will it have frequent minor improvements but never really get to the point of AGI (basically growing asymptotically)?