ChatGPT already mounts a challenge to Google. Today, it: 90%:"absolutely correctly and factually", 10%:"wildly misleadingly but plausibly" succinctly answ

ChatGPT already mounts a challenge to Google. Today, it: 90%:"absolutely correctly and factually", 10%:"wildly misleadingly but plausibly" succinctly answers specific queries nearly instantly on every subject, usually saving a lot of time and giving satisfying answers, but leaving you wondering "did it that just make that up though?" The 10% of the time it makes stuff up and teaches it to you feels like a very mean-spirited practical joke. It is at once the most informative and useful tool, able to tutor you in reams of real-world information in minutes, and also the most blatantly dis-informative resource, to be constantly on edge with to see when it is pulling your leg. By combining it with Bing's results, it could check to see if what it is hallucinating has any basis in reality. By removing the 10% of times it just makes things up out of whole cloth that are not anywhere on the Internet, users can use it without wondering "Perfect answer! But I wonder if it just made that up."