Doubling down on comments he’s made throughout the years regarding AI’s potential impact on the medical industry, legendary Silicon Valley investor and Sun Microsystems founder Vinod Khosla said on Wednesday that he believes “any radiologist to plans to practice in ten years will be killing patients every day” because machine-powered solutions will have advanced to such a point that they’ll be far more effective than professional human practitioners.
Speaking at the closing keynote of Creative Destruction Lab’s Super Session in Toronto, Khosla also said on stage that “radiologists are toast,” and that they flat out “shouldn’t be a job,” continuing that in a decade when AI-based diagnostic technology has advanced, people in this profession will “be causing deaths, because [they] choose to practice.”
The position was in keeping with his past statements on the subject, dating back to as early as 2017, when he expressed the belief that some types of doctors would be “obsolete” within five years (the timeline seems to have gotten a bit longer in the interim, but he qualified later that this include the time it will take for acceptance of the fact that the tech is better by the community and general public). Khosla added that he also believes that oncologists will also be surpassed by alternatives based on domain-specific AI solutions, but that that’s probably a bit further out on the 15-year horizon.
Instead, he believes that human general practitioner doctors will be more valuable, and will work with AI solutions for more specialized medical fields often currently considered more highly skilled. This is in keeping with the general thinking about how narrowly focused AI is easier to accomplish than machine intelligence that addresses more general topics.
Khosla noted further that oncology is “much easier to automate” than the job of a factory worker, since the job of a factory worker “has much more dimensionality.”
The investor qualified the strength of his statements by adding that he believes the time for being polite is over, since he does believe that on balance people will be more dangerous than machine intelligence in the specific domain of radiology in the ten year timeframe.