Uncovering User Affect Towards AI in Cancer Diagnostics
Despite the rapid application of artificial intelligence (AI) to healthcare, we know comparatively little about how users perceive and evaluate these tools. Following “dual route” theories of information processing from decision science, we propose that because users lack the expertise to rationally understand AI through cognitive evaluation, they rely on their feelings or heuristic route processing to make judgments about AI systems and recommenders. Therefore, affect becomes an important component that influences people’s willingness to adopt AI—and this may be especially true in a context like personal health, where affect is both explicit and heightened. Using the context of remote dermatological skin cancer screening, we examined people’s affective perceptions of an autonomous AI algorithm capable of making recommendations about skin lesions (as either cancerous or benign). In a three-stage study (n = 250), we found that people do hold complex affective responses toward AI diagnostics, even without directly interacting with AI. Findings are relevant to designers of AI systems who might consider how users’ a priori affect may make them more or less resistant to technological adoption. Additionally, the methodological approach validated in this study may be used by other scholars who wish to measure user affect in future research.
KeywordsAffect Healthcare Audience Artificial intelligence
This work was supported by the National Science Foundation (Award No. NSF 1520723). The authors thank Rachelle Prince for her help with data collection.
- 4.Chang, S.F., Harper, M., Terveen, L.G.: Crowd-based personalized natural language explanations for recommendations. In: Proceedings of the 10th ACM Conference on Recommender Systems, RecSys 2016, pp. 175–182. ACM, New York (2016). https://doi.org/10.1145/2959100.2959153
- 8.Eslami, M., et al.: First i “like” it, then i hide it: folk theories of social feeds. In: Proceedings SIGCHI Conference on Human Factors in Computing Systems (CHI 2016), pp. 2371–2382. ACM, New York (2016). https://doi.org/10.1145/2858036.2858494
- 11.Haenssle, H., et al.: Man against machine: diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists. Ann. Oncol. (2018). https://doi.org/10.1093/annonc/mdy166
- 13.Johnson, C.Y.: The tech industry thinks it’s about to disrupt health care. Don’t count on it. https://www.washingtonpost.com/news/wonk/wp/2018/02/09/health-care-the-industry-thats-both-begging-for-disruption-and-virtually-disruption-proof/?utm_term=.c7b4312afdd7. Accessed 12 Dec 2018
- 14.Katz, D., Price, B.A., Holland, S., Dalton, N.S.: Data, data everywhere, and still too hard to link: Insights from user interactions with diabetes apps. In: Proceedings of 2018 CHI Conference on Human Factors in Computing Systems. ACM, New York (2018). https://doi.org/10.1145/3173574.3174077
- 15.Kouki, P., Schaffer, J., Pujara, J., O’Donovan, J., Getoor, L.: User preferences for hybrid explanations. In: Proceedings of the 11th ACM Conference on Recommender Systems (RecSys 2017), pp. 84–88. ACM, New York (2017). https://doi.org/10.1145/3109859.3109915
- 17.Lee, M., Baykal, S.: Algorithmic mediation in group decisions: fairness perceptions of algorithmically mediated vs. discussion-based social division. In: Proceedings of SIGCHI Conference on Computer Supported Cooperative Work (CSCW 2017), pp. 1035–1048. ACM, New York (2017). https://doi.org/10.1145/2998181.2998230
- 18.Loewenstein, G., Lerner, J.S.: The role of affect in decision making. In: Davidson, R.J., Scherer, K.R., Goldsmith, H.H. (eds.) Handbook of Affective Science, pp. 619–642. Oxford University Press, Oxford (2003)Google Scholar
- 22.Schwarz, N.: Feelings-as-information theory. In: Van Lange, P., Kruglanski, A., Higgins, E.T. (eds.) Handbook of Theories of Social Psychology, pp. 289–308. Sage, Thousand Oaks (2011)Google Scholar
- 25.Ventola, C.L.: Mobile devices and apps for health care professionals: uses and benefits. Pharm. Ther. 39(5), 356–364 (2014)Google Scholar