New Orleans A five-week nurses’ strike recently ended with hospitals in New York City. There have been other strikes around the country led by the National Nurses Union, SEIU, and other unions. Casual observers may read the headlines about wages and patient care, nod, and keep going, believing it’s same ‘ol, same ‘ol. Reading a recent examination in Scientific American on the bumpy rollout of artificial intelligence in hospitals, makes it pretty clear that there is a lot more going on with these job actions. Nurses have become the frontline in holding AI accountable in hospitals.
These stories are not for the weak of heart. Even for this staid and careful scientific journal, you know something is wrong, when the subtitle of the article is “AI is spreading through American medicine. When the system is wrong, the burden lands on those who never asked for a copilot in the first place.” Yep, they are clear the first and last line of defense is the floor and specialist nurses.
They lead with a story of sepsis, a stubborn killer prompted by infection, and a major cause of death in US hospitals. Many hospitals, including the one in Nevada, where this case was based, have bought an artificial-intelligence system, which issues a warning when AI senses on coming sepsis. The nurse attending the patient noticed she had a dialysis catheter, which would mean that her kidney’s wouldn’t be able to keep up, if she were flooded with IV fluids that would then end up in her lungs. The charge nurse “told him to do it anyway because of the sepsis alert generated by the hospital’s artificial-intelligence system.” He refused, saving the patient based on his experience and individual analysis of the patient’s need. But what stayed with the nurse was the waterfall of events prompted by the AI-alert. “A screen prompted urgency, which a protocol turned into an order; a bedside objection grounded in clinical reasoning landed,…as defiance. No one was acting in bad faith. Still, the tool pushed them to comply when the evidence right in front of them – the patient and her compromised kidneys – demanded the exact opposite.”
See what I mean? This is scary stuff. The article was full of similar examples and concerns expressed by nurses on the front line. Several hospital level union leaders were quoted about their fears that AI was essentially just nowhere near primetime, and absolutely not a substitute for nurses and doctors at the bedside.
The hospitals universally demurred when contacted, claiming their use of AI was simply as another “tool” in the healthcare arsenal, but their comments ran as contradictions to nurses’ own experiences. In fact, some of the company’s producing and selling these AI solutions, admitted that they were still making improvements and newer versions were better. Gulp, that’s a bit frightening as well, making patients and their hospitals, the guinea pigs on this AI learning curve between life and death. Some hospitals had abandoned the AI systems after a year’s trial. Some systems they bought, believing they would save time and hours, turned out to be duds or only produced marginal advantages.
Hospitals are rationalizing buying and deploying these AI systems to produce more efficient care and deal with persistent staffing shortages. Nonetheless, it is impossible to read all of this and not come away with a better understanding of why nurses are demanding more staffing and better patient ratios, and the money to allow them to continue to ply their professions. I started reading the piece as a supporter already, but the more I understood their stories and digested the lame rationalizations of the hospitals, the more I was wishing there might be a general nurses strike everywhere to protect us all, whenever we’re forced to go into a hospital.