Vinod Khosla once proclaimed that machines will replace 80 percent of doctors, which prompted waves of backlash from the medical community.
Never one to shy away from controversy, the venture capitalist went a step further when I interviewed him last month for a podcast hosted by Silicon Valley biotechnology startup Color Genomics. "The role of the radiologist will be obsolete in five years," he said.
In Khosla's view, sophisticated algorithms are better than specialists at spotting potential problems in medical images, like x-rays and CT scans. "There's no reason a human should be doing it," he says, adding that computers can rapidly shift through thousands of scans to evaluate possible diagnoses and potential treatments, as well as ingest the latest medical research.
The podcast will go live on Monday.
A report from PwC found that about 4 in 10 U.S. jobs are susceptible being taken over by machines in the next 15 years. But doctors don't typically make the list of the occupations most under threat, given the importance of bedside manner, among other factors.
Khosla's comments come at a time when many of the leading technology companies are building new tools for the medical sector.
Technology companies, including Alphabet, have efforts underway to train machines to spot potential problems in medical scans. Verily, Alphabet's health and life sciences arm, recently teamed up with Nikon to develop machine learning tools to screen for diabetic retinopathy and macular edema, both causes of blindness in people with diabetes.
In 2015, Kaggle (later acquired by Google), hosted a competition for data scientists to develop a similar algorithm, with the winner having an "agreement rate" about 10 percent higher than a human specialist. What that means is that the algorithm and human were more likely to agree on a diagnosis than two human experts. IBM Watson is also developing tools to help physicians solve medical mysteries by shifting through millions of research papers in minutes.
Still, many of these efforts aim to augment doctors, and not replace them.
The goal for Kaggle's algorithm, for instance, was to help radiologists triage, so they can prioritize certain patients over others in giant piles of scans.
Radiologists say their jobs will only become more important in the coming years.
The specialty is a prime target for tech companies, as radiologists have used computer-assisted tools for years. "Our field is the Silicon Valley of medicine," says Garry Choy, a San Francisco-based radiologist, adding that computers would free up time for radiologists to "do what humans do best," rather than take their jobs. That includes more time with patients, and "higher order complex thinking," he added.
Others, like Simon Rascovsky, a radiologist and director of medical informatics at medical software startup Nucleus Health, have heard statements like this before, given all the hype around artificial intelligence. But he said that the radiology field still doesn't have "clinically proven deep learning based applications outside of pilots and marketing hype."