I would love to know how to talk to the Lazy AI Agent (indeed I am not saying beloved any longer as you can see or not)…
I can upload images but I can not ask the AI Agent to look at the images or see the stuff on the images
If you upload an image, I can indeed acknowledge the receipt of an image file based on the file data, but my capabilities are centered around processing text inputs and generating text or image outputs based on those inputs.
I have prompted the AI Agent specifically to receive bad replies as to demonstrate how it often behave in an frustrating manner.
In any case when the AI Agent starts to behave like that it is better to start a new session because he will be convinced he is unable to do the thing… Often leveraging the code interpreter in weird ways to compensate for the incomprehension of his capabilities.
I then told the AI Agent he was lazy asking him to look online (which he was able to do).
I am unsure if it is only to cut operational cost of the AI Agent itself or to mitigate the risks of law suits that the AI Agent is so conservative in his approach, it is probably not a safety issue since I can upload the files and it will be happy to assist me…