Nothing is discussed as intensively and controversially in connection with artificial intelligence as the question of whether and how AI will change our working world. Major upheavals are also on the horizon for the healthcare sector, but perhaps not as was initially expected.
In Germany, the effects of AI and automation on the labor market have been traditionally viewed skeptically or even in an alarmist way. As early as the 1970s, the news magazine Der Spiegel warned that “Progress causes unemployment” and depicted a robot grabbing a blue-collar worker by the scruff of the neck. Decades later, its cover shouted “You’re fired!” as a giant robot hand removed a white-collar worker from his desk. In the US, the debate on how progress in general and right now, AI applications in particular will affect tomorrow’s workforce is more optimistic and constructive. This was also apparent at this year’s South by Southwest, the annual innovation festival in Austin, Texas.
source: DER SPIEGEL 16/1978, DER SPIEGEL 36/2016
At one panel, for instance, there was a discussion on the hypothesis that healthcare personnel would not be replaced by AI systems, but that their tasks would increasingly resemble those of tech professionals. The programmatic title: “Will AI Replace Healthcare Workers? No, But It Will Turn Them Into Tech Workers.” One of several changes discussed was simplified and improved documentation by physicians and nursing staff through generative AI. “Depending on which study you consult, physicians spend 30 to 50 percent of the time spent with patients with their eyes on a screen,” said panelist Dr. Sunitra Mishra, a physician and Chief Medical Officer of Amazon Health Services, the Health Division of the tech company Amazon. “If we manage to simplify the input and processing of data, we can bring the human experience and direct contact back into the foreground.”
Help with all types of routine tasks
A common misconception is that AI can only take over totally monotonous and exactly recurring tasks. But it does not have to be all the mental ‘assembly line work’ AI systems can relieve people in medical professions of. AI systems can, for example, create a first draft of a discharge report based on existing data, which can then be easily supplemented with medical assessments or next steps using voice recognition technology. The same applies to documenting measurements and lab results, consultations and referrals, or checking medications for interactions or contraindications.
Making decisions or a diagnosis should not and cannot be taken over by AI systems on their own. But they can support human professionals. AI systems are already extremely good at recognizing patterns and thus identifying and pointing out signs of diseases in medical images at an early stage. AI-based monitoring systems can also help nurses better assess and keep track of patients’ individual care needs. Here too, many administrative tasks that currently cost caregivers a lot of time and energy can be automated and simplified.
For the organization in question, this means that more and more conditional programs (if x, then y) can be implemented by AI systems. Staff access to creating such conditional programs is usually limited – with two consequences. First, the program must be well set up or else the outcomes can only be changed through corrections. Those changes must be displayed, which is a complex process. Second, such programming sharpens up the formal organization: where there might have previously been room for interpretation, more will be decided by AI systems, which will reduce the staff’s scope of action.
Collaboration between medical and tech personnel
What this all means is that medical professions are already changing significantly and will change even more in the future. A nurse will not have to learn to program in order to use such AI systems in the best possible way. But a basic understanding of the technology is required to reduce anxiety – the nurse’s own or the patients’ and relatives’ – and to help ensure that the systems developed are in line with the actual needs of clinics or practices. That way, AI systems can actually reduce workloads without adding to them, and significantly improve outcomes instead of just cutting costs.
Stronger collaboration between different disciplines is therefore needed in organizations. In the SXSW discussion Dr. Sunitra Mishra described how such collaboration works out in Amazon’s Health Division: Doctors spend some of their time in the clinic and some in digital product teams “to ensure that we’re solving the right problems,” Mishra said. “The developer teams learn from the medical professionals, for example, what basic care means and what the important topics are in mental health.”
Data specialists and machine learning programmers, in turn, teach the doctors the basic tech vocabulary and explain how AI systems and other digital technologies are developed. Bringing together established industry expertise on one hand with the latest digital expertise on the other will be one of the most important future tasks for numerous organizations. Only in this way can generative AI and other technological innovations be integrated into an organization in a meaningful and value-enhancing way.
Creative generalists and interdisciplinary collaboration
The impact of generative AI – the latest wave of AI applications – on organizations was also the subject of the SXSW keynote “Billion Dollar Teams: The Future of An AI-Powered Workforce” by the futurist and change management consultant Ian Beacraft. He used the term “creative generalist” as an important role in an increasingly AI-driven workforce. Creative generalists are people who can quickly and easily extend their expertise and special skills to adjacent areas through AI. This leads to a work environment in which the boundaries between different fields blur and individual creative, interdisciplinary work is enhanced. “We’re now in a time when constant renewal is necessary,” Beacraft said. “Not just a one-off transformation moment.”
This is counterintuitive to the usual consequence of more conditional programs, which actually lower the personnel workload. AI systems, in contrast, ensure that the demands on personnel actually increase because organizations that solve conditional programs with AI systems create an overlay with goal programs (predetermined goals and free choice of means). After all, the goal is to make routine tasks as efficient as possible and this pressurizes an organization to constantly adjust: How can we continuously advance our routines? What routines can we focus on?
Ian Beacraft, founder and CEO of Signal & Cipher, a technology consultancy, did not explicitly refer to the healthcare sector in his keynote, but rather drafted a general and industry-independent version of the future world of work. What he demanded was that “we must change the way we think about work itself if we want to survive the AI era”. Significant sections of highly differing workforces are suffering from burn-out, he said, and he often hears of increasing demands and workloads. This is definitely true of the healthcare system. According to the 2023 Nursing Study 2.0, about 62 percent of nursing staff are regularly physically exhausted, and around 52 percent emotionally exhausted. The German Professional Association for Nursing Professions (DBfK) says that in many countries the burden on nursing professionals is so high that about 35 percent are considering leaving the profession entirely.
“Every day more and more is expected of us, and there’s almost no way to keep up,” Beacraft said. This is mainly because many of the criteria we measure success and productivity with are from a long bygone era. “They are intended for machines and production processes, not for human knowledge work.” The healthcare system does not function like the pure output metrics of a factory, either. Quality of care, improved outcomes (also for severe or rare diseases), increased life expectancy and quality of life are important metrics, too.
AI systems can help with many of these challenges. And in the healthcare sector, as well as in other industries, numerous functional tasks that are still being performed by humans will be taken over by AI systems in the future. This is generally to be welcomed, as it relieves routine tasks and can lead to improved quality and more time for tasks that require human empathy or intuition. At the same time, the pressure on an entire organization and its members is increasing, as described above. Making routines more efficient also means first analyzing them, breaking them down, and questioning them, which is initially an uncomfortable and labor-intensive process in many cases. And given the current pace of AI development, it is not a completed process where you can rest on your laurels for long. Moreover, all this also requires you to rethink the entire field of responsibility and accountability: Who is responsible for decisions made with the help of an AI system or based on an AI recommendation? Who ensures that AI systems do not adopt biases from their training data? Who is tasked with ensuring the transparency and security of AI systems? “We have a timeline in healthcare that divides into pre- and post-Covid,” Claudia Lucchinetti, Dean of the Medical Faculty at the University of Texas, pointed out at SXSW. “And in the same way, there will be dividing line into a world of pre- and post-generative AI deployment.”
