I’ve been doing technical interviews for decades and have gotten good feedback from candidates about my technique. I have a small number of questions that are nearly impossible to have prepared for - one for instance is finding the median in a large dataset using map reduce. The question is open enough there’s a lot of solutions that are expensive, a lot that are dead ends, and one that’s pretty clever. The key is I let them investigate different possible solutions and give them guidance as they butt against expensive or impractical solutions and guide them to the answers I’ve heard over the years that work best. Because I ask the same questions over and over again I can tell how people treat the problem and try to solve it. But since it’s not a typical interview question no one is prepared to answer it off the cuff. Ultimately my goal is to assess whether the candidate can understand the problem, understand dead ends when I point it out, understand how to pivot to another approach, and understand how to interpret hints in a positive direction. IMO these are the important skills for a professional problem solver working in a team.

I wouldn’t use automated screening tools. These tools select for irrelevant criteria in my opinion, and are mostly useful if as an interviewer you’re too lazy or too incompetent to formulate your own questions that tell you whether a person can solve problems and interact well in a group problem solving situation. But I’m an oldster so YMMV.