Human preferences toward algorithmic advice in a word association task

Abstract

Algorithms provide recommendations to human decision makers across a variety of task domains. For many problems, humans will rely on algorithmic advice to make their choices and at times will even show complacency. In other cases, humans are mistrustful of algorithmic advice, or will hold algorithms to higher standards of performance. Given the increasing use of algorithms to support creative work such as text generation and brainstorming, it is important to understand how humans will respond to algorithms in those scenarios—will they show appreciation or aversion? This study tests the effects of algorithmic advice for a word association task, the remote associates test (RAT). The RAT task is an established instrument for testing critical and creative thinking with respect to multiple word association. We conducted a preregistered online experiment (154 participants, 2772 observations) to investigate whether humans had stronger reactions to algorithmic or crowd advice when completing multiple instances of the RAT. We used an experimental format in which subjects see a question, answer the question, then receive advice and answer the question a second time. Advice was provided in multiple formats, with advice varying in quality and questions varying in difficulty. We found that individuals receiving algorithmic advice changed their responses 13$$\%$$ % more frequently ($$\chi ^{2} = 59.06$$ χ 2 = 59.06 , $$p < 0.001$$ p < 0.001 ) and reported greater confidence in their final solutions. However, individuals receiving algorithmic advice also were 13$$\%$$ % less likely to identify the correct solution ($$\chi ^{2} = 58.79$$ χ 2 = 58.79 , $$p < 0.001$$ p < 0.001 ). This study highlights both the promises and pitfalls of leveraging algorithms to support creative work.

Explainability does not mitigate the negative impact of incorrect AI advice in a personnel selection task

“…Correct advice was followed in approximately half of their decisions and incorrect advice in about a tenth. Other studies have found similar effects 17 , 19 , indicating that people often follow advice regardless of whether it is correct or not 9 , 10 , 16 . Participants’ tendency towards overreliance may be attributed to the advice serving as a decision anchor.…”

“…On the other hand, it has also been shown that individuals are more willing to adhere to algorithmic than human advice ( algorithmic appreciation 15 ). This tendency is particularly evident in tasks with measurable outcomes that require logical problem-solving or in judgments under uncertainty 16 . The underlying mechanisms leading to either algorithmic aversion or appreciation are not yet fully understood, with several advice characteristics proposed as potentially relevant, including the quality of advice.…”

Evidence of a cognitive bias in the quantification of COVID-19 with CT: an artificial intelligence randomised clinical trial

“…AI inconsistencies in underserved patient populations are a known issue 22 . Automation bias 23 , 24 is another related pitfall that roots in the overreliance of the radiologists in a computerized system. Regardless of this potential risk, the trial’s main conclusions remain unaltered, by design, as the endpoint is objectively measured on the final reports, directly.…”

Algorithms in selection decisions: Effective, but unappreciated

“…In fact, they found evidence for "algorithmic appreciation," where individuals preferred algorithmic advice over human advice, including their own judgments, when compared directly. The inclination to favor algorithms was observed in several "objective" areas, such as logic problems (Dijkstra et al, 1998), word association task (Bogert et al, 2022), assessing the magnitude of visual stimuli (Logg et al, 2019), predicting product demand (Daschner & Obermaier, 2022), and choosing lottery numbers (Farjam, 2019). It was also evident in taste-related contexts, such as ranking the popularity of songs, predicting romantic attraction (Logg et al, 2019), and divulging personal information and feelings (Lucas et al, 2014).…”