Artificial Intelligence And The Scientific Method: How To Cope With A Complete Oxymoron

W. Clark Lambert,Muriel W Lambert,Mohammad Hassan Emamian,Michał Woźniak,Andrzej Grzybowski
DOI: https://doi.org/10.1016/j.clindermatol.2023.12.021
IF: 2.797
2024-01-12
Clinics in Dermatology
Abstract:Artificial intelligence can be a powerful tool for data analysis, but it can also mislead investigators due in part to a fundamental difference between classic data analysis and data analysis using AI. A more or less limited data set is analyzed in classic data analysis, and a hypothesis is generated. That hypothesis is then tested using a separate data set, and the data are examined again. The premise was either accepted or rejected with a value "p," indicating that any difference observed was due merely to chance. In contrast, a new hypothesis is generated in artificial intelligence as each datum is added to the data set. We explore this discrepancy and suggest means to overcome it.
dermatology
What problem does this paper attempt to address?