The age of machine learning (ML) and artificial intelligence (AI) is poised to dramatically alter the practice of medicine in many disciplines, including the field of liver disease. How the radiologist diagnoses hepatic tumors and how the pathologist interprets liver histology are among the areas in which the digital revolution is being incorporated in gastroenterology and hepatology. However, as others have opined, several challenges in the technology involved in deploying these various ML/AI applications remain to be overcome. [1,2] Attempts to computerize the diagnosis of druginduced liver injury (DILI) have been on the minds of hepatologists for years but have thus far remained elusive. In this issue of Hepatology, Hayashi and colleagues, [3] all of whom have long-standing expertise in the causality assessment of DILI, present their evidencebased, computerized modification of the Roussel Uclaf Causality Assessment Methodology (RUCAM), which they have renamed Revised Electronic Causality Assessment Method (RECAM), for diagnosing DILI. Their attempt to bring causality assessment into the digital age is a welcome addition to our diagnostic armamentarium. But although their offering has several advantages over the current version of RUCAM, as the authors themselves recognize, as with other attempts to enter the computerized age of medicine, it remains a work in progress.Given the hundreds of drugs, herbal products, and chemicals that can cause liver injury and mimic every possible form of acute, chronic, benign, and malignant liver disease, the ability to make a firm diagnosis of DILI has remained clinically challenging in the absence of a definitive diagnostic biomarker. Many health care professionals, especially nonhepatologists, rely heavily on consultation with an individual with expertise in the field of drug hepatotoxicity. But despite more than three decades of research searching for a validated and specific biomarker, DILI remains a diagnosis of exclusion. [4] Efforts to devise semiobjective algorithmic approaches to analyze the important elements needed to suspect possible DILI-i.e., a compatible time to onset (latency), time to recovery after the drug has been stopped (positive dechallenge), and the ability to adequately exclude alternative causes-have given rise to a number of causality assessment methodologies over the last