From memory I think this takes place in Schild's Ladder by Greg Egan, but in real time. People have personal AI’s wired into their brains that can be asked to converse with others.
I'm always interested by the relationship between science fiction ideas, how the idea changes as it becomes near-term... and how the idea manifests in real life.
AIs wired into brains is a complex endpoint. It's a plausible nexus of technologies or culture, a fit premise for a plausible fiction. Fold space, spice monopoly and galactic aristocracy. As sf approaches near term, the nexus tends to simplicity. Once we get to imminent, the nexus can get downright subtle. At this point, the agents of technological change might be a catalyst or process... not necessarily a breakthrough. The breakthroughs may have already happened. It's just a little hard to know, until after the affect.
The point we're at now, plausibly, has us on the precipice of "AI avatars having a discussion for us." That path from here to there, is relatively banal. Autocomplete, autoreply. It could happen with or without moments of decision "I hereby decide to devolve such-and-such powers to my ai avatar because Y." Consequences can be profound, but decision making rarely is. The banality of will.
FWIW, I can really see this happening quick and hard. In fact, now is the first time I've had non-abstract or distant concerns about ai. The proliferation of a gpt enhanced software keyboard has some unpredictable potentialities.
There's also Phoenix Exultant by John C Wright, where there are AIs that are basically clones of you that are empowered to make all sorts of decisions for you. IMO it's more approachable sci-fi, but that's not always what folks are looking for.
