Since the recent model updates, which were meant to improve the use of real-time web data for current events, I’ve been experiencing a persistent and frustrating issue where ChatGPT (GPT-4) often gets “stuck in a loop” when handling questions about current events.
Here’s the pattern: When I ask an initial question on a specific current event, the model provides a summary-style response that blends multiple sources. However, if I then try to follow up for clarification or ask a more specific detail, it frequently just repeats the same response instead of addressing the follow-up.
Example:
I asked, “Has Trump nominated more cabinet appointees as President-elect than normal?” The model responded with a general answer referencing other administrations, like Biden’s, and mentioned something like, “Biden had nominated [X number of roles] by May.”
I then tried to clarify, asking, “At this point in time post-election, is Trump’s nomination pace actually faster than previous presidents?” The model just repeated the same answer about Biden and “May,” without providing any new information specific to the question.
Issue: This isn’t a one-off error but seems to be a consistent pattern. The model frequently defaults to looping back to its initial answer when a follow-up question is asked on a specific detail.
Is this a recognized issue or bug? Are there any fixes in the pipeline to help GPT-4 handle follow-up questions on current events more effectively?