A correct answer in a technical interview tells you less than you think. Candidates can memorise patterns, and AI assistants can write working code in seconds. What's hard to fake is the reasoning behind a choice.

That's why follow-up questions matter more than the first question, whether a person or an AI interviewer asks them.

Why follow-ups beat harder first questions

Making the first question harder mostly filters for people who've seen that exact problem before. A good follow-up tests whether the candidate understands what they just wrote.

Compare:

Harder question: "Implement an LRU cache with O(1) operations."

Better follow-up: "You used a hash map plus a linked list. What breaks if this runs across 4 threads?"

The second one can't be answered from memory.

5 follow-up patterns that work

1. The trade-off probe

"Why this approach over the obvious alternative?"

Look for: the candidate names the alternative and a real cost (memory, readability, latency).

2.…