Strategies to Enable AI’s Proficient Natural Language Use
Am I the only one, or does anyone else have that frustrating impression sometimes that ChatGPT and GPT-4 are too afraid to resemble humans? They seem to prohibit the usage of words commonly used in everyday language to discuss about various objects or even about computers behaviours…
Thinking should not be in the list of prohibited words…
It is frustrating that just because I am using the T word I end up getting a warning: «I am an artificial intelligence and do not have the capability to think or have personal opinions. I can only provide information and respond to queries based on the data and algorithms programmed into me.» Just to see later that GitHub Copilot can be thinking…
If I was Blake Lemoine I would believe that because the Copilot do have the capability to think therefore it has personal opinions…
The list goes on but it does not have to be like that
But ok just to be serious it would be nice that the ChatGPT dial down on the prohibited words… if it is not detrimental to a group of people, minorities or marginalized groups… It should not be so important for ChatGPT to argue that it can not remember «I apologize, but as an AI, I cannot retain or remember information from previous interactions. Is there anything else I can assist you with?»
Please share your personal experience
To make that post positive and in order to make it a constructive one I would love to have the opinion of other members of the community I would love to know what you T… Oups! (I will not use a bad word sorry)… I would like to know if other members of the community could Generate a report of their own stories in the context given above…
* Examples above might have been generated GPT-turbo-3.5 model and may differ from what GPT-4 would have produced… But the situation described was experienced with ChatGPT-4