The rise of AI as a hot workplace topic generates both excitement and trepidation for workers. It is also easy to understand why the hype prompts mixed feelings. Such advances can improve productivity by performing rote tasks, thereby freeing humans up for more human concerns. Then comes the anxiety, as workers understandably worry about being replaced by robots.
That fear weighs heavily upon current strikes in California, but on a more (literally) vital note, the healthcare industry is grappling with how far to dive into the ether. Putting patient health into AI hands could prove to be a risky business.
How much of that risk is acceptable, and how can hospitals tiptoe into testing AI while also adhering to the Hippocratic Oath of doing no harm?
Those questions are being batted around, particularly in research hospitals, yet the healthcare AI revolution has ultimately begun. And after telemedicine revamped healthcare during the pandemic, AI could reshape healthcare much more drastically.
Whether one is talking about a chatbot or an algorithm, important considerations exist:
- AI’s immediate benefits: AI promises to increase operational efficiency and can flag patients at risk for deadly conditions like sepsis or predict those likely to suffer a fall. AI can also interpret radiology tests with stunning accuracy. Other early successes include whole-body MRI scans for cancer. At Mount Sinai, AI is test-monitoring critical care patients and generating risk scores based on vital signs. Algorithms can then dig through databases of conditions to match combinations of symptoms with possible diagnoses. AI can also organize doctors’ clinical notes.
- AI could ease the nursing crisis: Dozens of strikes took overworked nurses off the job over the past few years. Could their burdens measurably be lessened by AI? A further question from workers is whether executives’ primary goal of such tech would be helping the bottom line rather than improving care.
- No substitute for humans: No software can entirely replace a nurse’s physical observations of patients, such as an unusual odor or erratic behavior. One nurse recently made the case that AI is prone to misdiagnosis, which has led her into “moral distress” over knowing that her expertise could be secondary to an algorithm.
Another harbinger of uncertainty exists. A recent survey indicated that around 75% of healthcare execs are willing to embark upon the AI journey, but “only 6% have an established generative AI strategy.” The work has only begun for AI in this industry, and this is also a realm where upgrades will be needed regularly.
Finally, healthcare employers must also consider that the feds are dragging their feet on how to regulate AI, so one must be prepared for on-the-fly retrofitting. That headache doesn’t even touch the ethical implications of allowing AI to make decisions about human life.
With all of that said, this train has left the healthcare station. Now, to decide how far to let it ride.
