AI Trained on Old Scientific Papers Makes Discoveries Humans Missed (vice.com)

I call bullshit on this one: "the algorithm was able to provide candidates for future thermoelectric materials, some of which may be better than those we currently use". Unless any of these novel candidates was tested and proven to have desired thermoelectric properties, no discovery was made, contrary to the hype in this title.

Check my algorithm, which just proposed a steaming hot pile of male cow excrement as a candidate for thermoelectric material. Test it while it's still hot!