We have an intern that is finishing a four year computer science degree that has no clue what git is, never used a log and all he presents is AI garbage.
I find it profoundly depressing to try and teach someone who has no interest in the craft.
80% of the candidate I interview pass (leetcode style coding interview, as mandated by the company). This is actually annoying because I'll probably have to raise the bar and start rejecting very good candidates.
I'm sorry but to me this part reads like a humorous phrase that's popular in some circles in my region which goes:
"Maybe <list of negative things, usually correct characterizations of the speaker>, but at least <something even worse>"
The companies I worked for used automated coding quizzes like Codility to weed out the worst applicants, but I suspect you're already doing that.
How is them knowing when binary search is useful relevant to what they'll be doing at work should they get hired?
Because of our work is changing, faster than ever, not day to day but over time. You need a foundation to handle that change. My 2X years experience showed me that the people who has strong foundation handle the transition well. If I'm going to hire and invest and mentor, I want that person to be successful.
Because it goes directly to their understanding rather than whatever rote memorization they’ve done. Anything that involves rote memorization can be done, better, by LLMs. What’s in short supply are people with good critical thinking skills and the ability to deal effectively with new problems.