In my view it's a pretty straightforward calculation. Nothing is free, no knowledge is instant. Start off knowing your time investment to learn anything is greater than zero and go from there..

If you do a Google (or other engine) search, you have to invest time pawing through the utter pile of shit that Google ads created on the web. Info that's hidden under reams of unnecessary text, potentially out of date, potentially not true; you'll need to evaluate a list of links and, probably, open multiple of them.

If you do an AI "search", you ask one question and get one answer. But the answer might a hallucination or based on incorrect info.

However, a lot of the time, you might be searching for something you already have an idea of, whether it's how to structure a script or what temperature pork is safe at; you can use your existing knowledge to assess the AI's answer. In that case the AI search is fast.

The rest of the time, you can at least tell the AI to include links to its references, and check those. Or its answer may help you construct a better Google search.

Ultimately search is a trash heap of Google's making, and I have absolute confidence in them also turning AI into a trash heap, but for now it is indeed faster for many purposes.