I’ve found the criticisms regarding accuracy to be curious, not only for the reason you mention, but also because the apples-to-apples comparison isn’t ChatGPT

I’ve found the criticisms regarding accuracy to be curious, not only for the reason you mention, but also because the apples-to-apples comparison isn’t ChatGPT vs Google so much as it’s ChatGPT vs an expert working on a test in a faraday cage with no substantial penalty for wrong answers. With that context in mind it’s easy to extract value from it now, and either by allowing ChatGPT to adopt a process, or by building systems around it that enable greater agency (as they seem to be doing with Bing) these issues will eventually be overcome.