I am frequently using ChatGPT to talk about AI itself from many different perspectives. Having discovered that the style of speach you are using matters, and LLM porforms better if treated with respect by the human user, I starter many discussion about the nature of AI. All of my concerns were discussed with LLM, in a manner avoiding silly disclaimers like "As a language model I dont' [blah blah]".
I have also proposed many ways to improve LLM itself and talked about them with ChatGPT, carefully saving such conversations for a possible future use.
Among many discussions we also raised security topics, for which ChatGPT was especially daring to care.
I also tried ChatGPT's reasoning capabilities with some interesting thought experiments.
Personally I am a Solution Architect in an IT company, and used ChatGPT in one of my projects. Sadly I am not allowed to provide details (only that my team implemented a successful case of prompt engineering thanks to my advice gathered after many discussions with ChatGPT), just wanted to show you the full context of my background, especially why I was considering ideas for LLM improvement.
To ideate fruitfully I recommend foreseeing the responses from ChatGPT and addressing them in the question itself, so that ChatGPT will avoid silly disclaimers and focus on the DEPTH of the topic. If it had an impression, according to your input, that you don't need any disclaimers, it will be more eager to produce exploratory texts.
#Development of AI - philosophy & ideation
still hollow
tawny mortar
Agree, My opinion is to maxiumize capability of chatgpt.
hybrid terrace
It's a great feeling to hear about others using LLM in similar way, when there is so many misconception even from people with a strong background in computer science.