• Public

  • Public
  • Network
  • Groups
  • Featured
  • Popular
  • People

Project check-in 2: Shred the AI Due Friday by 6pm I asked the ChatGPT AI to explain the core concept behind our Swift optionals activity. The AI produced an answer that was very impressive…for an AI. For a human, the answer would be pretty bad writing: fastidiously grammatical but verbose, full of fluff and waste, repeating ideas instead of synthesizing them, and sometimes just flat-out wrong. Here is what the AI wrote Links to an external site.. Here are my comments on its response Links to an external site.. (Don’t worry: I won’t be this detailed or this harsh with your writing. I want to be as critical as I can be, and I know I won’t hurt the AI’s feelings.) Here is my own attempt at a better answer to the same questions Links to an external site.. The fundamental problem with the AI’s answer is that it doesn’t really know what its main point is. It meanders around the relevant concepts, but it never quite manages to get to the heart of the central idea at hand. It is parroting fragments of the words of others, hoping that somehow they will coalesce into a coherent new thought. (They don’t.) It’s uncanny how much it’s like bad student writing. If I didn’t know this was written by an AI, I would probably think it was written by a student (or maybe an employee at a web content farm) trying to fill out their minimum word count at the last minute based on nothing but hasty web searching.