News Summary

• If you're not careful, these chatbots can steer you completely off course when searching for an answer.
• I found that probing its logic with more questions was a good way to get the AI to snap out of it.
• When you do this, you force the LLM-based chatbot to check its own work before making a claim.
• Luckily for me, I can recognize when the AI might be hallucinating, since I know enough about the subject matter to see that the information it relays could be false.