A new academic field called algorithmic accountability has grown from the question of big data accuracy. More experts want to be able to see the routes big data programs are using to collect and synthesize information, according to the New York Times.
IBM is working on a solution for algorithmic accountability for its Watson platform. With the assistance of researchers at a group of healthcare organizations including Cleveland Clinic, Rochester, Minn.-based Mayo Clinic and Memorial Sloan Kettering Cancer Center in New York City, IBM is developing a program called Watson Paths to allow physicians to see the inference paths Watson used to reach its conclusions, according to the report.
Watson has the potential to aid providers in clinical decision making with its ability to draw from millions of medical documents every minute, collecting data points much faster than many people ever could. However, the accountability software will make it more palatable and trustworthy to people because the logic is explained, according to the report.
Other companies are outsourcing algorithmic accuracy oversight. One company, San Francisco-based technology-enabled lending company Earnest, has a staff member review the predictive recommendations of its lending software just in case of errors. If the staff member spots an error, he or she can veto the algorithm’s decision, but it must be based on evidence rather than a gut feeling or preference related to the algorithm’s decision, according to the report.
