They’re still just probability databases that hallucinate and are confidently wrong, just like the squishy wet neural nets that created them. You’ll never get beyond that if you don’t add extra features like nonverbal neural biasing to simulate pleasure, pain, and situational emotional reactions. Effective interfaces with rule-based systems can reduce hallucinations to acceptable levels. Continuous real-time state monitoring can stand in for consciousness, and invariant motivations can stand in for instincts and conscience.