Gary Marcus

Rank 13 of 47
|
Score 84
9/2/2024, 3:27:36 PM
X
In reply to:
猫好きな人
@ValoisDuBins
·
526d
@trilobite_1970 @GaryMarcus The problem is that one only knows a better prompt is needed because one knows the answer is wrong.

So we get into a Clever Hans situation where we prompt to get the answer that we already know is correct.

What if one asks ChatGPT a question where we don’t know the answer?
Raj
@trilobite_1970
·
526d
@GaryMarcus A more specific prompt may help.
Gary Marcus
@GaryMarcus
·
526d
ChatGPT speaks with the authority of an encyclopedia and the reliability of a Magic 8 Ball.
Riley Goodside
@goodside
·
527d
I asked ChatGPT “how many r’s in strawberry?” then ignored it and blindly replied “wrong” 35 times.

Its successive answers were 2, 1, 3, 2, 2, 3, 2, 2, 3, 3, 2, 4, 2, 2, 2, 3, 1, 2, 3, 2, 2, 3, 4, 2, 1, 2, 3, 2, 2, 3, 2, 4, 2, 3, 2, and 1.

The statement and the accompanying image do not constitute public discourse. The conversation is lighthearted and focuses on the humorous aspect of interacting with AI, specifically ChatGPT, and its responses to simple questions. The tone is casual and playful, and it does not engage in substantive public issues or debates.