To the Editor:
We were interested to see that our recent editorial entitled “A Crack at MAC,”1 was classified as relating to crack cocaine. In particular, the editorial was indexed on the Anesthesiology website (https://pubs.asahq.org/anesthesiology/article/134/6/835/115687/A-Crack-at-MAC) with the topics “crack cocaine” and “minimum alveolar concentration” (fig. 1). We feel obliged to make a disclaimer: Potential readers with an interest in addiction medicine will be disappointed. We presume that the classification links are generated by some sort of artificial intelligence classification process. This misclassification demonstrates the profound limitations of the semantic depth of artificial intelligence processes. These processes can link patterns nicely, but, underneath all the frothy hubris, artificial intelligence is a zombie that would see nothing wrong with classifying the iconic Pink Floyd song (“Another Brick in the Wall”) as a masonry manual. By chance, a survey on Artificial Intelligence in Anesthesiology has just been initiated (“2021 Survey on Artificial Intelligence in Anesthesiology,” Carlos Estrada Alamo, M.D., M.B.A., Penn Medicine, University of Pennsylvania Health System, Philadelphia, Pennsylvania). Could you trust a machine that can’t even understand that the word “crack” has at least 11 different meanings, and might even be a verb?
Editor’s note: This error has been corrected and the classification identification system adjusted to prevent future mistakes of this type.
Competing Interests
The authors declare no competing interests.