
Im getting quite annoyed with this model. Its bad at following instructions. In this chat it generated vidoes twice when i didnt ask it too. I mostly use claude otherwise. What i was asking for in this chat was kinda ridiculous, but i still think the model is overly eager for my taste. Getting a bad impression of tool calling capabilities, even though this its just one example and not enough to make conclusion.
Credentials and full chat here for context:
M.Sc. in Applied AI. Senior AI Advisor / Data Scientist. Started working as a ML researcher in 2019. Have processed around 200 000 conversions with claude haiku in varying data pipelines. Building two AI companies and have been doing prompt engineering workshops regularly since early 2023.
