This is something I have been pondering for some time in the context of on line safety and now playing with ChatGPT and developments in deep fake technology has prompted me to clarify my thoughts.

ChatGPT is based on neural nets which have been in development for some decades. the inner workings of modern neural net algorithms is very complex and in use obscure. Some applications of Neural nets technology, for image recognition and medical diagnosis have been very successful.

There is a lot more to ponder on the idea of a digital personal assistant, particularly for the elderly. Trust is paramount and is distinctly lacking in the current state of AI.