[SUPPORT] Example 2: Question and Answering

[SeaDude]

trafficstars

Hello,

https://github.com/openai/openai-cookbook/blob/main/examples/gpt4o/introduction_to_gpt4o.ipynb

I appreciate how you didn’t edit or cherry-pick the results from Example 2. Because that’s what we see in the wild.

What are some approaches to prevent or deal with the hallucinations shown in Example 2: Question and Answering?

Do we just build the workflows and wait for the models to get better in the future?

Do we somehow post-process the response e.g. use other LLMs to validate, always keep human in the loop, etc?

A cookbook example of these methods would be very valuable.

Thank you