Are any of ChatGPT's statements erroneous? (how come Terence Tao gets this high quality output but I get hallucinations when I ask simpler math stuff?)