Experts warn that more research is needed on the behavioral impact of artificial intelligence technologies after studies found they could put people at increased risk. .
The researchers found that the placebo effect appears to be at work when people believe AI technology is improving their performance. We found that people with high expectations of these performance technologies took more risk in their decision making.
The team behind the findings says more research is needed to help people better understand how these technologies affect behavior.
“People who believe they are enhanced by cutting-edge technologies such as AI and brain-computer interfaces are more likely to take risks,” said Robin Welsh, an assistant professor at Finland’s Aalto University.
“This happens even when no real enhancement technology is involved. This suggests that this is due to people’s expectations rather than a noticeable improvement. It suggests that a strong belief in improvement based on
The researchers used a well-established psychological experiment, the Columbia Card Task, to measure participants’ levels of risk-taking when they thought AI was being used. This involves participants gaining or losing points for revealing cards with hidden values. This group of 27 participants was led to believe that a form of AI was being used to improve cognitive performance.
However, participants did not know that the game was cheating and that performance techniques provided little benefit. Nonetheless, most of the group believed AI was helping them perform better, resulting in more risky decisions.
Steven Villa, Postdoctoral Fellow at LMU Munich, said: It can lead people to riskier decisions and favorable user ratings, which can have real consequences. “
With more and more areas employing AI, such as using vision-enhancing technology to help firefighters see through smoke, researchers are optimistic that biased expectations will lead to greater risks. Concerns have been expressed that the desire to
Professor Thomas Kosch of the University of Berlin said: “AI-based technologies that empower users are becoming increasingly commonplace, playing a role in real-world decision-making that impacts people’s lives, well-being, confidence and safety. Placebo-controlled studies for accurate evaluation and validation to distinguish snake oil from genuine innovation are needed to ensure its efficacy.”
