Researchers Chart Alarming Decline in ChatGPT Response Quality

For example, Chat GPT-4 prime number identification accuracy fell from 97.6% to 2.4% from March to June 2023.

2 comments

I have noticed the decline. Current ChatGPT is making significantly more factual errors now than it did in the past.
Oh good, so, 3.5 did improve. I thought it seemed better at doing what I asked of it. I barely use GPT-4 these days