We have been experimenting with chatbots for our use case since about 2008 and now with LLMs they are getting close to what we have needed all along. It is nice to have access to the currents LLMs for general knowledge and for the other stuff that they can now do but we HAVE to be able to generate Avatars that have as much of the real-life memories, feelings and attitudes (I know, I know) for a single, likely older, biological human person.
Some of us have a lot of text, still and video imaging information about ourselves and I have been looking at tech that might convert this information into stuff that could be useful for what we want - but the main problem remains - do people have suggestions about how we might process this information that would be continuously updated from the real world by the individual - with this “Avatar” ALSO having access to one or other LLMs for all the other stuff that they can do?
If I can get started with the Avatar for myself - even it means experiments with various approaches, then once we settle on something that seems workable and information portable - for future-proofing for use with other tech that might follow LLMs for example - then I could make the service available to our members here:
If anyone is motivated they could also comment here: