Is ChatGPT any more factually accurate with its results than Microsoft Copilot when chatting with these AI?

I was chatting with Microsoft Copilot this morning in the hope of understanding why the Democrats did not fight for extension of the enhanced subsidies for the ACA with a filibuster in the Senate in early March when they had the opportunity before the big beautiful bill was passed by Congress and signed by the President on July 4.

Based on this chat I conclude that this version of AI is far from ready for prime time as I will explain below.

I found that Copilot consistently made factual errors that only improved after instructing it to focus in on only the most reliable source for Senate votes which is the Senate.gov website. Copilot stated that it produces results from several sources on the internet which led to the factual errors (Senate vote record) it acknowledged it had been making.

The errors it had been making weren't small. It listed former Senators in its results for certain continuing resolutions and stated that certain Senators voted yes instead of no and vice versa in the data tables it created that effectively were hallucinations until I told it to use only the Senate.gov website. After that, the results appeared to be accurate.

Am I expecting too much of AI or not? It seems to me that the vast amount of money being spent on new AI data centers is not to achieve AGI but instead to get AI accurate enough to depend upon, especially when it should be verifying its results for factual accuracy before spitting them out. Not too much to ask for imo.

It seems I need to be extremely specific with my queries to get accurate factual data from Copilot and know which sources to instruct it to focus on to hope to get trustworthy results.

After this morning's Copilot chat session, I no longer am for anything that requires it at this time.

So. does ChatGPT show the same tendency of generating factual errors as Copilot?