ChatGPT's Confidence Can Conceal Incorrect Answers
The hosts warn that ChatGPT can sound equally authoritative when it is right or wrong. They also identify stale information and weaker performance on obscure long-tail queries as important limitations that preserve part of Google's advantage.
- Confident language makes errors difficult to detect
- The discussed version used information from before 2021
- Obscure topics appeared to expose weaknesses
- Google remains difficult to disrupt because many searches are novel
“it is wrong as much as it is right, but you can't tell the difference”
“It doesn't appear to be as good as a very obscure topics.”