In the last message, I described how ChatGPT 3.5 confidently gave a wrong answer to my prompt "compare igpu in i3-1005G1 and i3-1215U". Now I'm trying free Google Gemini. New this month! It's answer is much better. But that makes its serious mistakes more difficult to catch. First of all, I blame Intel. It really doesn't describe the differences in iGPUs when showing the difference between the processors: <https://ark.intel.com/content/www/us/en/ark/compare.html?productIds=196588,226269> Gemini incorrectly says the i3-1005g1 supports DirectX 12.1 whereas Intel says 12. Gemini says the i3-1005g1 supports OpenGL 4.5 whereas Intel says 4.6. Gemini says that the i3-1215u's iGPU has 80 Execution Units whereas Intel says 64. I'll stop here. Gemini has wasted my time. If I were less careful, it would have mislead me. This is the comparison that I trust the most (but it could be wrong too) <https://www.notebookcheck.net/i3-1215U-vs-i3-1005G1_14098_11413.247596.0.html>