This blog is long overdue, but I will try to retrospectively summarise most of the research and thinking I’ve been doing over the past few years, as well as document thoughts and plans going forward.
One of the big questions of this research is “Can an avatar act?”
Who better to attempt to attempt an answer at this point in the history of Artificial Intelligence than the one and only ChatGPT.

If you are referring to the concept of an “Avatar” as a digital representation of a person or character in a virtual world, then the answer is that it depends on how you define “acting.”
An Avatar can certainly be programmed to perform a variety of actions and movements within a virtual environment, such as walking, running, jumping, or waving. However, whether or not you consider these actions to be “acting” would depend on your perspective.
In general, when we talk about acting, we are usually referring to the portrayal of a character’s thoughts, emotions, and motivations through their behaviour and dialogue. While an Avatar could be programmed to perform scripted actions or deliver pre-recorded lines of dialogue, it would not be capable of improvising, reacting to other performers, or conveying the full range of human emotions in the way that a trained actor could.
So, while an Avatar can certainly perform actions and interact with its virtual environment, whether or not you consider this to be “acting” would depend on your definition of the term.
Hmm. Many of the latest Machine Learning techniques, as exemplified by ChatGPT, are indeed effectively simulating a type of improvisation. So, in order to provide these systems with the possibility of being able to “act” in the way it is described above, perhaps all we need to do is provide enough training data so that the model would know how and what to perform in any “given circumstance”, while suitably constrained by the predicted behaviour and habituations of the character they are tasked to “create”.