This past Wednesday an article came out that showed evidence that Watson, while be able to win at Jeopardy and predict the weather, it would not be a viable candidate for process improvement for the medical industry. Watson recommended to a patient with cancer and severe bleeding, chemotherapy and a drug that could cause sever hemorrhaging. An IBM spokesperson did comment on Watson’s track record which has “helped with 13 different cancer treatments in 230 hospitals for 84,000 patients”. Maybe we chalk that one misdiagnosis to “US Cities”?
Do you think Watson should be allowed to recommend diagnoses even though it clearly will make mistakes once in awhile? Should further safeguards be implemented before Watson could recommend specific medications? When recommendations are made, should Watson be required to also display the logic behind those recommendations?