NEW YORK – Artificial intelligence is currently linked to specific problem solving and is not some form of Terminator model capable of handling multiple tasks with autonomy. In other words, each time you hear the term “AI,” it is a computer solving a specific problem or task using algorithms “and not ‘thinking’ like you and me,” said Ido Weinberg, MD, assistant professor, Harvard Medical School, Boston.
Dr. Ido Weinberg
AI is present in daily life – everything from cellphones to the Alexa voice interface on a smart speaker. That AI system, however, is amassing data, learning about you, and using that data intelligently, Dr. Weinberg said at a symposium on vascular and endovascular issues sponsored by the Cleveland Clinic Foundation.
AI in health care make sense, he said, because the health sector is a vast consumer market with potential for financial gain. Repetition, which is common in the health sector, is one of the foundations required for using AI and robotics. If a task can be repeated, then it means a machine can do it, said Dr. Weinberg.
The spread of AI and robotics one day may improve health care accessibility in remote areas where physicians with the appropriate training may not be available.
AI is already at work in the health care industry. “Pulmonary nodule detection can be done better with machines than by people, pathological identification and scanning of various slides can be done better by a machine than by a humans,” he said.
Artificial intelligence also can be designed to detect emotion by assessing various cues in phrasing, key words, and tone. These AI functions already are being used by sales reps on the phone to defuse and control interactions with customers and complainants. AI also can be implemented in interactions with people, which is an important part of dealing with patients, Dr. Weinberg said. Drug discovery is a key area where AI is flourishing, as well.
Luckily, in terms of physicians keeping their jobs, there are barriers to the use of AI to replace clinicians, Dr. Weinberg pointed out. Health care is not a monolith, and every specialty is different, meaning AI would have to be tailored to each task and specialty for each unique field. Quick proliferation of AI across the board is unlikely, especially when the varying roles of nurses and physician assistants are included.
Although robots in science fiction stories and films often are capable of multitasking a variety of needs, robots at present are much more limited in real life. In surgical situations, for example, they can perform specifically tailored tasks but cannot extend beyond those defined parameters as a real surgeon can, according to Dr. Weinberg, and this lack of flexibility is a severe limitation on the expansion of AI into health care.
Despite these limitations, Dr. Weinberg urged attendees to consider how AI can be used to facilitate their work.
“Believe in the roadblocks, but be a fast adopter – an early adopter – and understand where AI can currently augment you and make you better and more productive,” he said. “And keep doing procedures; AI and robotics currently have a problem with most of those,” Dr. Weinberg concluded.
Dr. Weinberg reported no conflicts relevant to his talk.