Task-specific algorithm advice acceptance: A review and directions for future research

Kaufmann, Esther; Chacon, Alvaro; Kausel, Edgar E.; Herrera, Nicolas S.; Reyes, Tomás

doi:10.1016/j.dim.2023.100040

Cited by 10 publications

(1 citation statement)

References 77 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some evidence indicates that DMs are less inclined to accept, trust, or rely on algorithmic advice than human advisors, a phenomenon named "algorithm aversion" by Dietvorst et al (2015). In a recent review, Kaufmann et al (2023) found algorithm aversion to be present in 75% out of 122 tasks (44 studies) reviewed. People tend to reject algorithms in a variety of domains, such as medical, where algorithms are consistently trusted and used to a lesser extent than human doctors (Bigman & Gray, 2018;Longoni et al, 2019;Promberger & Baron, 2006), financial, when predicting prices of stocks (Önkal et al, 2009), admission decisions, (e.g., Dietvorst et al, 2015), educational evaluations (Kaufmann & Budescu, 2020) and also in contexts that reflect personal taste such as what books to read or movies to watch (Sinha & Swearingen, 2001), what jokes would be funnier (Yeomans et al, 2019) and in moral questions (Dietvorst & Bartels, 2022).…”

Section: Algorithms As Decision Aiding Toolsmentioning

confidence: 99%

Algorithms in selection decisions: Effective, but unappreciated

Rabinovitch,

Budescu,

Meyer

2024

Behavioral Decision Making

View full text Add to dashboard Cite

Selection decisions are often affected by irrelevant variables such as gender or race. People can discount this irrelevant information by adjusting their predictions accordingly, yet they fail to do so intuitively. In five online studies (N = 1077), participants were asked to make selection decisions in which the selection test was affected by irrelevant attributes. We examined whether in such decisions people are willing to be advised by algorithms, human advisors or prefer to decide without advice. We found that people fail to adjust for irrelevant information by themselves, and those who received advice from an algorithm or human advisor made better decisions. Interestingly, although most participants stated they prefer advice from human advisors, they tend to rely equally on algorithms in actual selection tasks. The sole exception is when they are forced to choose between an algorithm and a human advisor. In that case, they pick human advisors. We conclude that while algorithms may not be people's preferred source of advice in selection decisions, they are equally useful and can be implemented.

show abstract

Section: Algorithms As Decision Aiding Toolsmentioning

confidence: 99%

Algorithms in selection decisions: Effective, but unappreciated

Rabinovitch,

Budescu,

Meyer

2024

Behavioral Decision Making

View full text Add to dashboard Cite

show abstract

Do We Use Relatively Bad (Algorithmic) Advice? The Effects of Performance Feedback and Advice Representation on Advice Usage

Daschner,

Obermaier

2024

Behavioral Decision Making

View full text Add to dashboard Cite

Algorithms are capable of advising human decision‐makers in an increasing number of management accounting tasks such as business forecasts. Due to expected potential of these (intelligent) algorithms, there are growing research efforts to explore ways how to boost algorithmic advice usage in forecasting tasks. However, algorithmic advice can also be erroneous. Yet, the risk of using relatively bad advice is largely ignored in this research stream. Therefore, we conduct two online experiments to examine this risk of using relatively bad advice in a forecasting task. In Experiment 1, we examine the influence of performance feedback (revealing previous relative advice quality) and source of advice on advice usage in business forecasts. The results indicate that the provision of performance feedback increases subsequent advice usage but also the usage of subsequent relatively bad advice. In Experiment 2, we investigate whether advice representation, that is, displaying forecast intervals instead of a point estimate, helps to calibrate advice usage towards relative advice quality. The results suggest that advice representation might be a potential countermeasure to the usage of relatively bad advice. However, the effect of this antidote weakens when forecast intervals become less informative.

show abstract