You’re in the park going for a run, and your wearable device is tracking your performance, your heart rate, and various other aspects of your physical health. Pooling this data over a period of time gives you a strong idea about your physical fitness. Combine that data with your diet, your genetic data, your electronic medical records and you can paint a comprehensive picture of your physical health.
Making sense of this data, together with any symptoms you volunteer, is increasingly the preserve of autonomous technology that can absorb vast quantities of data at a time when doctors report inputting data into electronic medical records as a key source of stress. Would you trust the diagnoses of such autonomous systems or would you prefer to have a human doctor have the ultimate say in the recommendations you receive?
That was the question posed by new research from Penn State, who attempted to understand the way people interact with the latest generation of medtech tools.
The analysis revealed, perhaps unsurprisingly, that the comfort people had with technology more generally, both in terms of their confidence in the capabilities of the technology and their ability to utilize it, had a big impact on their willingness to accept the diagnoses of autonomous doctors.
“There is increasing use of automated systems in the medical field, where intake is now often conducted through a kiosk instead of by a receptionist,” the researchers explain. “We investigated user acceptance of these ‘robot receptionists,’ along with automated nurses and doctors. In addition, we tested whether the form that these roles took—human-like, avatar or robot—made a difference in user acceptance.”
Automated assistance
Healthcare is well known for its adherence to Baumol’s disease, which dictates that costs remain high because the labor costs associated with delivering healthcare have remained high for generations. While technology has advanced, it hasn’t really led to a reduction in headcount in hospitals, and so costs have spiralled ever upwards.
The Penn team highlight the tremendous bandwidth issues afflicting most doctors today, with burnout common across the industry. They believe that AI-based technologies can do a lot of work for doctors, and because they can crunch huge quantities of data 24/7 without needing breaks or holidays, they can provide invaluable assistance to a profession in dire need of support.
It should be said that the volunteers were recruited from Amazon’s Mechanical Turk, which must surely have resulted in a more tech savvy audience than is perhaps the norm. Each volunteer provided their preconceived beliefs about technology, and especially AI-driven technologies. Of particular interest to the researchers was something known as machine heuristics, or in lay terms, the stereotypes we have about machines, especially in terms of their efficiency and objectivity.
Adherence to such a machine heuristic was gauged by asking the volunteers to state their level of agreement with comments such as, “When machines, rather than humans, complete a task, the results are more accurate.” Alongside these questions the volunteers were also asked to rate their level of expertise using the latest technologies.
After these questionnaires were complete, the volunteers were presented with a range of healthcare-related scenarios where either the doctor, the nurse or the receptionist was portrayed by a human, a digital avatar or a machine. The aim was to test the acceptance level of each volunteer of the service given by the healthcare provider, and their willingness to use them in future.
“We found that the higher people’s beliefs were in the machine heuristic, the more positive their attitude was toward the agent and the greater their intention was to use the service in the future,” the authors explain. “We also found that power usage predicted acceptance of digital healthcare providers. A power user (a person with advanced computer skills) is more likely to accept a robot doctor, for example, than a non-power user.”
A double dose of technology
Interestingly, if people had high levels of technological ability and strongly supported the machine heuristic, they had the most positive attitude towards potential automation of healthcare provision. The authors contend that this is perhaps an ideal combination of abilities and mentality to underpin the acceptance of autonomous technology in healthcare.
This was largely consistent regardless of whether the volunteers were engaging with doctors or receptionists. If the volunteer were strong users of technology and had the high adherence to the machine heuristic, they were supportive of digital healthcare.
“Our results suggest that the key to implementing automation in healthcare facilities may be to design the interface so that it appeals to expert users who have a high belief in machine abilities,” the researchers conclude. “Designers can direct resources toward improving features such as chat functionality instead of anthropomorphizing healthcare robots. In addition, increasing the number of power users and the general belief that machines are trustworthy may increase the adoption of automated services.”
Interacting with machines
To navigate this kind of scenario, a team from the MIT Media Lab believe a whole new field of research is required. They make their case in a recently published paper, in which they argue for a multi-disciplinary approach to future technology development that takes into account the active role technology is playing in our lives. It demands that technology moves on from being seen as a passive participant in our life towards one where they’re an active player with their own behavioral characteristics.
“We’re seeing the rise of machines with agency, machines that are actors making decisions and taking actions autonomously,” the team explain in a blog. “This calls for a new field of scientific study that looks at them not solely as products of engineering and computer science, but additionally as a new class of actors with their own behavioral patterns and ecology.”
There has been a huge amount of time and energy devoted to understanding how humans can work effectively together. Now, as autonomous technology enters both the workplace and other parts of life, the time has come for this to be expanded to include how man and machine can combine forces productively.
