[sampo]

In a way, you have respect the integrity of the PhD student. But also, publishing a fraudulent conference paper in this field (neural networks), the real damage to anything is minimal.

Nobody expects a random paper in this field to be particularly true or useful. A lot of papers are trivial applications of old ideas, obfuscated behind complicated word soup. Or false good results due to buggy code, which you can never check because they don't publish the source code. Or useless or irreproducible due to many other reasons. The field produces also good results, but you need to follow bigger trends, as individual publications are mostly just noise.