Is there actually an effective way to handle queries with RAG where time periods are relevant? I made a proof-of-concept RAG for documents on a government website shortly after GPT-3.5 came out and remember this being a big problem. The most glaring wrong answer was "Who is currently the governor?" It answered with the previous governor, likely because he was listed as such in 8 years of documents versus the 2 years at the time for the current governor.