IBM’s artificial intelligence platform Watson was touted as a revolutionary tool for diagnosing and treating cancer. So far, Watson for Oncology has failed to deliver on those claims, struggling to learn about different types of cancer and only making recommendations based on training by its human operators, not its own insights.
Its shortcomings are profiled in an in-depth investigation published by STAT News, which interviewed dozens of people familiar with the system and examined its use and performance in hospitals around the world. What STAT found was, despite IBM’s claims that the “cognitive era” in healthcare has begun thanks to Watson, the system is struggling with the same issues of interoperability and data collection which dogs most health IT products. The hours spent feeding information into the system results in little payoff, with oncologists complaining it’s not directly helping in recommending treatments and isn’t providing any new insight.
“IBM ought to quit trying to cure cancer,” said Peter Greulich, a former IBM brand manager. “They turned the marketing engine loose without controlling how to build and construct a product.”
Read the full investigation at the link below: