The AI would sometimes go off on a tangent instead of asking conceptual questions (e.g. real-world applications of random things). However, I didn’t experience any significant level of inaccuracy like I sometimes do with other AI models, and it responded well and in-depth to my answers, often probing me for more information to test my understanding.