I totally agree. My thesis advisor was doing semantic search research, I was the one doing ML. I never understood the need to organize information when you can just throw ML at it. But the research might be a stepping stone in AI further down the line, when you need to build exhaustive knowledge instead of "intuitive" knowledge based on pattern matching. So who knows what it's good for.