My impression of chatgpt so far is that it's good at replicating existing human behavior, but bad at doing anything out-of-sample. It anchors quickly onto patterns that have been done by many people before, but there's definitely something that we call "original thinking" that it's not doing at all.

As a random example, consider these AI-generated debates (credit to pmarca for the format).

Notice that the Chinese perspective is not even consistent across the two debates! It's pattern matching to the usual yes/no debate, and... https://i.imgur.com/OebQ19c.jpg

Adding a few small changes ("in China we have regulations", "progress", "haves and have nots") to try to match up to each side's unique cultural features. The basic structure of the debates contains nothing that hasn't been said on the internet hundreds of times before.

In general, this makes it seem more likely than before that we'll have a pretty long period in which AI makes the world weird but before it takes away everyone's job or risks turning the world into paperclips.

Where do you think this model can provide true value? AI coding assistant seems super interesting.