Tired of COVID-19? Artificial intelligence can help, but we are sceptical
Can artificial intelligence (AI) relieve the strain on the healthcare sector? The technology exists, but the Danes are very sceptical of it according to a new study from Aarhus BSS at Aarhus University. Currently, only one in ten Danes would prefer a medical AI to a human physician.
Would you also like to return to a normal life without PCR or rapid tests, without working from home, and without face masks and hand disinfectant? Then you are far from the only one.
However, are you ready to explore new solutions to avoid future lockdowns and restrictions? For instance, by making use of new technological advances in artificial intelligence? If the answer is yes, you are one of the few.
Did you know that in fact, a lot of intelligent solutions already exist which could ease the pressure on straining healthcare sectors all around the world? For instance, an artificial intelligence has been developed to diagnose COVID-19 through recordings of people’s coughing, and this AI can be transferred to an app on a smartphone.
In general, the use of artificial intelligence (AI) in the healthcare sector could prove to a be a much less disruptive but just as effective measure in battling the current COVID-19 pandemic as the pervasive national lockdowns, according to a group of researchers that includes researchers from Aarhus BSS.
Red and green curves
Indeed, the risk of overloading hospitals has been the predominant argument for introducing national lockdowns as well as extensive and often very disruptive restrictions in the public as well as private sphere during the last two years of COVID-19. And right from the beginning of the epidemic, the Danes became quite familiar with the red and green curves presented by Minister for Health Magnus Heunicke: The number of hospitalised patients, which was at all costs to be kept below the number of available beds in the healthcare sector.
“If we can relieve physicians, nurses and other healthcare professionals in the often straining healthcare sectors of some of their workload, any form of help will be of great importance. And this is where AI could represent an alternative. The technology exists, but the question is how patients would choose if they could choose between different options,” explains postdoc Darius-Aurel Frank from the Department of Management at Aarhus BSS.
He has headed a group of researchers within consumer behaviour and conducted a study which was to examine people’s attitudes and willingness to accept medical artificial intelligence as an alternative to seeing a human physician. The study has been carried out in Denmark and France respectively, and during an ongoing pandemic, which makes it particularly interesting in relation to prior studies.
The study has been published in the journal Plos One and is a collaboration between researchers from Aarhus BSS, the University of Agder in Kristiansand, Norway, and TBS Business School in Toulouse, France.
“There is need of specific measures and information campaigns if the use of AI in the healthcare sector is to gain greater support and trust. But at the same time, you need to be very careful and balance communicative efforts so as not to create mistrust towards the existing healthcare sector and human physicians.”
Postdoc Darius-Aurel Frank, Department of Management, Aarhus BSS
And if you are a strong advocate of the use of AI within the healthcare sector, the result is probably a little discouraging: Nine out of ten respondents prefer to be examined and diagnosed by the familiar figure of a human physician in a white lab coat and with a stethoscope around the neck compared to Doctor AI.
“The main obstacle to being diagnosed by an artificial intelligence is quite simply the lack of trust in AI. Despite the fact that there is no situation in which AI does not outperform people," says Darius-Aurel Frank.
The study also shows that there is no difference between Danish and French patients. They are almost equally suspicious of medical AI, and this is a surprise to the researchers.
”Denmark is the most digitised country in the world and is way ahead of France. For this reason, you would think that the Danes would be less negative towards medical AI than people in France. But apparently, this is not the case, which suggests that it takes more than a modernisation of public services to build trust in medical AI,” Darius-Aurel Frank points out.
However, the study also demonstrates that several aspects may impact whether medical AI is perceived as something positive or not.
Previous studies have shown that an important factor in people’s mistrust towards AI has to do with their perception that Doctor AI will not sufficiently take into account the special characteristics, circumstances and symptoms which apply to their individual situation. This is called uniqueness neglect.
Interestingly enough, this study shows that fear of being subjected to uniqueness neglect from a human physician affects your inclination to perceive medical AI as something positive. But not the other way around.
“If you experience or fear that your physician or nurse does not pay sufficient attention to your personal, medical characteristics, it might push you towards a more positive attitude to medical AI,” says Darius-Aurel Frank.
This is true for people who feel disconnected from other human beings as well. They are more prone to choose a medical AI than a human physician, the study shows.
But perhaps the road ahead is not quite as long as one would believe at first. The study also indicates that if you can achieve a mere one unit increase in the measure of perceived trust in medical AI, the likelihood of people choosing medical AI over a human physician will multiply by seven.
According to Darius-Aurel Frank however, there exists a lot of misinformation about AI in general, and for this reason, he emphasises that any attempt to increase trust in medical AI is a delicate matter and has to be performed with caution.
“There is need of specific measures and information campaigns if the use of AI in the healthcare sector is to gain greater support and trust. But at the same time, you need to be very careful and balance communicative efforts so as not to create mistrust towards the existing healthcare sector and human physicians. This is not a zero sum game. AI should assist, relieve and help people, not replace them,” says Darius-Aurel Frank.
Facts about the study:
About the study: The present research investigates people’s intention to adopt medical AI as well as the drivers of this adoption in a representative survey of two European countries (Denmark and France, N = 1068) during the initial phase of the COVID-19 pandemic. Due to the rigorous control for confounds and reliance on domain-specific knowledge for inference of causality, the underlying cross-sectional data can be used to ensure external validity of the findings. That said, the replicability of this research may be limited due to the unique background of a global pandemic.
External collaborators: Tobias Otterbring, Department of Management, University of Agder, Kristiansand, Norway, and Institute of Retail Economics, Stockholm, Sweden. Sylvie Borau, Department of Marketing, TBS Business School, Toulouse, France.
External funding: The French data of this study were partly funded by the ANR-Labex IAST (under the Investissement d’Avenir Programme, grant number ANR-17-EURE-0010). The Danish data of this study were partly funded by a grant from the Aarhus University Research Foundation (Aarhus Universitets Forskningsfond; AUFF, grant number 28207) awarded to Tobias Otterbring.
Conflicts of interest: None
Link to the research article: doi.org/10.1371/journal.pone.0259928
What the researchers did:
The experiment was conducted among a total of 1,168 participants in France (602) and Denmark (566) respectively during the initial phase of the COVID-19 pandemic in April-May 2020.
The participants completed identical surveys on their attitudes and willingness to be examined and diagnosed for COVID-19 by medical AI compared to a human physician.
The participants were asked about their trust in a medical AI and a human physician respectively on a scale from 1-7, their trust in medical advice from a medical AI and a human physician respectively, and their trust in whether a medical AI and a human physician respectively would overlook the unique conditions of the participants, partly during the examination and partly during the subsequent processing of their test results.
Finally, the participants were asked to rank their support for the authorities’ anti-COVID-19 policies, their belief in conspiracy theories, their open-mindedness, their trait optimism, social belonging, self esteem, risk perception, and political ideology.