Yesterday, I wrote about some of my experiences using the OpenAI agent, ChatGPT. It’s an interesting piece of tech, but imperfect, joyless, and in some ways dehumanizing. It’s unclear what it might portend.
There’s speculation that ChatGPT could write plausible scientific abstracts, outputs difficult to distinguish from human-authored ones. From there, it doesn’t take much to believe that entire papers might someday be generated by tossing a results set into ChatGPT and directing it to write a scientific paper from it.
ChatGPT seems to relish the idea, as we’ll see below.