AI researchers allege that machine learning is alchemy

It feels like this is a natural consequence of the field gaining popularity very rapidly. A lot of unexperienced people come to it and simply apply off the shelf algorithms and methods to new domains in order to publish without understanding the details of the tools.

Not that long ago people where putting graphite or fullerenes into almost everything and publishing articles about “graphene/carbon nanotubes enhanced xyz” as well without doing basic verifications that the additions even made any difference at all, but still publishing because of the hype.

Is there a link to the paper?

Hmmm. I guess there really is no free lunch after all.