[trq_]

Definitely, but if you can detect when you might be in one of those states, you could reflect to see exactly which state you're in.

So far this has mostly been done using Reinforcement Learning, but catching it and doing it inference seems like it could be interesting to explore. And much more approachable for open source, only the big ML labs can do this sort of RL.

[TZubiri]

Right. The uncertainty will be high when responding to garbage inputs and it will be distributed along many tokens.

If probability(sum(tokens[:5])) < 0.5: Respond("I'm sorry I don't quite understand what you mean.")