I can tell the LLM nature of ChatGPT’s congratulatory tone, but generally feel it has strong analytical value and compares and contrasts seemingly different things well.
I write film and literature essays and it’s really good at finding overlapping or contrasting themes between works, like westerns, Twin Peaks, X-files, and Star Trek it understood without prompting that they all dealt with different types of frontiers.
It is also (90%) good at understanding satire, irony, layered communication, where the words might be associated with one thing but is saying the opposite.
Gemini oth, seems confused a lot by this and the carnival psychic routine of piecing vague words together is a lot more obvious. It often times doesn’t understand jokes that say one thing and mean another, or uses a word associated with something else but is changed by the context. And it will latch onto a word or phrase I used a use it ubiquitously in every paragraph.
[link] [comments]