From drafting emails to driving a car to diagnosing illnesses, despite the growing prevalence of artificial intelligence in daily life, how people feel about AI remains muddy. One study shows general distrust and general negative attitudes towards AI, while another line shows most positive attitudes towards technology.
So, under what circumstances would people be more willing to trust AI?
New research by Associate Professor MIT Sloan Coworkers explore this question and identify conditions that support AI in a particular domain, not in other domains. Takeout: People are open to AI in areas where technology feels better than humans and If the task at hand does not require personalization.
Personalization dimensions are important, but often overlooked. “Just because people perceive AI technology as more capable than humans doesn't mean they'll adopt it,” Lou said. “They are also interested in personalization. If any of these conditions are not met, people feel disgusted with AI.”
Consider your abilities and personalization
A team of researchers from LU and Sun Yat-Sen University, Fudan University, and Shenzhen University proposed a framework for ability and personalization that assumes that when determining the application of AI, it focuses on two important aspects.
- AI's perceptual ability: Is AI perceived as more capable than humans in tasks? Can it work better than people?
- Recognizing the need for personalization: Do tasks require human sensitivity or a personalized approach?
To test this framework, researchers conducted a meta-analysis of over 160 studies on a variety of AI applications involving over 80,000 participants. All studies empirically tested and compared AI vs. human preferences in a particular context. For each study, the coder group assessed the perceptual ability of AI in its context and assessed the need for task personalization.
Consistent with the framework of competence and personalization, researchers found:
- AI appreciation occurs when AI is considered to be more capable than humans and The task does not require personalization. For example, when predicting the future sales of a particular product or estimating the artistic duration of painting, people tend to prefer AI predictions over human predictions.
- Aversion of AI occurs when personalization is required when AI is perceived as less competent than humans. Or both.
For example, when it comes to skin cancer diagnosis or film recommendations, AI has proven to be better than people, but better. However, people are reluctant to rely on AI because these tasks require personalization.
Also, running experiments with human participants may not require much personalization, but in this domain, AI has yet to prove competent, so once again people are showing aversion to AI.
AI Executive Academy
Directly on MIT Sloan
Sign up now
How to improve your organization's AI adoption
Takeout for managers to deploy AI initiatives in their organization: “AI developers and organizations need to pay attention to not only the capabilities of AI, but also their usage,” Lu said. “AI is only appreciated if it is perceived as being more capable and deemed more necessary to personalize.”
Additionally, developers and managers need to understand how different people in the same context branch out into how AI values are interpreted. Recruiters who hire for recruitment may believe that screening and ranking resumes are impersonal and that AI can do better than humans. Those applying for the job dislike using this same technology because they perceive this same technology very differently and consider the recruitment application very personal.
Given such perceptions, organizations seeking to use AI should prioritize contexts that feature low personalization. Alternatively, you can reduce awareness of the need for personalization. Consumers need to recognize their bias and assess whether AI capabilities outweigh personalization concerns.
“Maxing AI's potential means understanding when it's welcome and not,” Lu said. “Only by addressing both abilities and personalization can we move towards meaningful human collaboration.”
Lu's research colleagues were Professor Xin Qin, Associate Professor Chen Chen, PhD students Hansen Zhou and Xiaowei Dong, and postdoctoral researchers at Limei Cao at Sun Yat-Sen University. Assistant Professor Xiang Zhou at Shenzhen University. Professor East Pun at the University of Houdun.
Read next: These human abilities complement the shortcomings of AI
