The first probelm with the concept wthat came to my mind was- how t ostructure it so the user woudl not feel lost while using the installation and also how should I limit the interactions so the project would fit he scope of this module.
In order to get more used to working with AI, I have made a few tests using the web GPT, prompting it to act as an old windows assistant, however I found its repsonses too generic. Eventually I have researched styles of comedians and bookwriters it ocould impersonate to make it more interesting, which produced a few interesting results.
Eventually I ave moved to the integration of chatGPT and Unity, where I encountered two obstacles- the API is not as well documented as I though it would be and throuhg my research I have not found a clear way to use it. The second, not really an obstacle, was that this sercvice requires payment. I have considred an idea, where instead of using the AI in real time I would instead primpt it to write all the interactions in the web browser and then copy those into the application. I got stuck on this idea and could not take it anywhere so after somet ime i have decided to abandon the AI idea completely