I'm surprised too. You'd think tech people would understand what's going on. But of the prior replies to your comment 7 out of 8 seem dismissive. I'm in the "obvious unstoppable progress" camp but we seem to be a minority.
I guess maybe it isn't that obvious - I've read quite a lot in the area. People saying LLMs aren't very good are a bit like people long ago saying chess programs aren't very good. It was true but there was an inevitable advance as the hardware got better and then that led to enthusiasm to improve the software and computers became better than humans in a rather predictable way. It's driven in the end by hardware improvements. Whether the software is LLM or some other algo is kind of unimportant.
I guess maybe it isn't that obvious - I've read quite a lot in the area. People saying LLMs aren't very good are a bit like people long ago saying chess programs aren't very good. It was true but there was an inevitable advance as the hardware got better and then that led to enthusiasm to improve the software and computers became better than humans in a rather predictable way. It's driven in the end by hardware improvements. Whether the software is LLM or some other algo is kind of unimportant.