This actually isn't that rare or as dire as it sounds -- when people ask me Science Things that are out of my purview, i often say that my stem degree is mostly not about Knowing Stuff or being able to do infinite complicated math off the top of your head, its about knowing how to figure out how to do the math, how to derive the Knowing Stuff from first principles, and where to go looking for and how to understand the information that you need to have to do that.

how to conduct an effective web search

I was also given courses that could be described as this; it was about how to find scientific journals, which ones to trust and which ones not to, when and how to contact a researcher directly instead of going through their company/journal/department, how to access papers 'legitimately' and for free, how to find and ethically use digital information that is permissible in peer reviewed papers, navigating the guts of NASA's old websites, how to scrape data, etc etc. Like, yes, sure, i suppose that that is just a very detailed and specific set of web-searches, but also, they absolutely would have had similar stuff in the pre-internet era, but it was just about library use, archival systems, and getting data physically posted to you instead.

These days, to not predicate modern science on access to the internet/worldwide data systems/computers would be like not assuming you'll have access to a fucking microscope or a library. Like yeah sure, at some point you might not, but i guarantee that then you will have much bigger problems. Scientists don't just know stuff, because that would require knowing almost everything in your field and that's not humanly possible. We know how to find it out, either through figuring it out or looking it up

EDIT: i'm not saying that this ai shit isn't shockingly and aggressively stupid, i'm just trying to make it clear that what your friend was talking about 10 years ago is almost certainly nothing like what's happening now. This is new and insane.