AI chatbots tell users what they want to hear, and that’s problematic
- Thread starter JournalBot
- Start date
From the Article:
I absolutely love this article for this sentence even if it's under-emphasized (in my opinion, that is).
Any answer given by AI LLMs is not given because it has confidence in being right/correct nor does it have a rational justifying the assertion it gives.
An LLM answer is given because the model statistics return that's the answer you're most likely to accept as real. It's not giving you answers, it's stringing together words you will likely ACCEPT.
Worse still, this implies that our own ignorance, biases, and presumptions are by definition the things we're most likely to affirm to an LLM as a 'good answer'.
Surely nobody would be foolish enough to rely on such things for thoughtful fact-based insight.............
I absolutely love this article for this sentence even if it's under-emphasized (in my opinion, that is).
Any answer given by AI LLMs is not given because it has confidence in being right/correct nor does it have a rational justifying the assertion it gives.
An LLM answer is given because the model statistics return that's the answer you're most likely to accept as real. It's not giving you answers, it's stringing together words you will likely ACCEPT.
Worse still, this implies that our own ignorance, biases, and presumptions are by definition the things we're most likely to affirm to an LLM as a 'good answer'.
Surely nobody would be foolish enough to rely on such things for thoughtful fact-based insight.............
Upvote
34
(34
/
0)
- Status