I think something like clip but that works for videos and transcriptions is where things start to get scary. At that point you would have a world model with ph

I think something like clip but that works for videos and transcriptions is where things start to get scary. At that point you would have a world model with physics and dynamics involved. As well as human expression and reaction.