The Hall of AI Fears and Hopes: Comparing the Views of AI Influencers and those of Members of the U.S. Public Through an Interactive Platform

TL;DR

The paper investigates whether AI influencers' views align with the U.S. public by building an interactive platform to collect public fears and hopes and assembling a 100-person Time100 AI influencer dataset. It then applies embeddings, clustering, and alignment metrics (including the $Q$-score and the misalignment score $\bar{\text{misalignment\_score}}_s$) to compare perspectives across three phases: influencer data curation, public data collection, and cross-group analysis. The findings reveal that the public emphasizes control and job-related fears, while influencers stress regulation and broader benefits, with notable misalignment across age, gender, and ethnicity; younger influencers and academics tend to align more with public views, while some underrepresented influencer groups diverge from their public counterparts. By delivering a publicly available anonymized dataset and a co-designed visualization platform, the work provides a concrete methodology and empirical insights to inform AI governance and inclusive design.

Abstract

AI development is shaped by academics and industry leaders - let us call them ``influencers'' - but it is unclear how their views align with those of the public. To address this gap, we developed an interactive platform that served as a data collection tool for exploring public views on AI, including their fears, hopes, and overall sense of hopefulness. We made the platform available to 330 participants representative of the U.S. population in terms of age, sex, ethnicity, and political leaning, and compared their views with those of 100 AI influencers identified by Time magazine. The public fears AI getting out of control, while influencers emphasize regulation, seemingly to deflect attention from their alleged focus on monetizing AI's potential. Interestingly, the views of AI influencers from underrepresented groups such as women and people of color often differ from the views of underrepresented groups in the public.

Figures (10)

• Figure 1: Overview of our three-step method for comparing the public's views and AI influencers' views on AI using an interactive platform. In the first step, we collected views from AI influencers--Time100 nominees ai100Time--by analyzing their interviews, focusing on key excerpts about AI's uses and impacts, and categorizing them into hopes and fears. In the second step, we built a platform to collect views from the broader public, starting with a literature review and interviews with two designers to elicit four design requirements and create the initial prototype (V1). Next, we engaged individuals from the public to iteratively update one design requirement and co-design three iterations of the platform (V2-V4). Next, we conducted a pilot study to evaluate the platform and identify improvements for the large-scale study. Finally, we used the platform to gather views from 330 U.S. individuals, representative of the population by age, sex, ethnicity, and political affiliation, capturing their fears and hopes about AI. In the third step, we compared the collected views.
• Figure 2: The "The Hall of AI Fears and Hopes" is an interactive platform consisting of six sections for collecting public views on AI and presenting the views of AI influencers. In the first section (A), participants share their definitions of AI. In the second section (B), they rate their fears and hopes, describe their specific fears and hopes, and indicate how widely they believe these views are shared. In the third section (C), participants can view portraits of AI influencers, organized by their more negative (left) and more positive (right) attitudes toward AI. By hovering (C1) and clicking on a portrait (C2), more information about the influencers' fears and hopes appears in a pop-up box. Participants proceed to the next section (D) by clicking on (C3), where they provide data on gender, ethnicity, age, education level, occupation, and AI literacy. In the fifth section (E), participants imagine a future 10 years ahead and vote on how to allocate money based on 10 randomly selected pairs of AI influencers' fears and hopes. In the final, sixth screen (F), participants see their symbolic portrait displayed alongside AI influencers, positioned based on their fear and hope ratings, as well as the similarity of their fears, hopes, and demographics (age, gender, ethnicity) to those of the influencers.
• Figure 3: Self-reported AI definitions across subgroups in our sample of the U.S. public. The chart reconstructs these definitions from left to right, highlighting popular adjectives ("human", "artificial", "smart"), nouns ("computer", "machine", "technology", "program"), and verbs ("learning", "answering", "creating"). The most common definition across participants is: "AI is a human-like computer capable of learning" (path marked in red).
• Figure 4: Average hopefulness scores across 20 subgroups based on age, sex, ethnicity, political affiliation, AI literacy, and job training and education requirements. The value "0" indicates being equally hopeful and fearful. Negative values indicate being more fearful, while positive values suggest being more hopeful. Standard deviations are shown as ± values next to the subgroup scores. The most notable differences in hopefulness are between older vs. younger participants, males vs. females, Democrats vs. Republicans, individuals with low vs. high AI literacy, and those in occupations requiring extensive training and education vs. those requiring less.
• Figure 5: Misalignment scores between subgroups of AI influencers and participants representative of the U.S. population. Young influencers' views are most closely aligned with those of our participants, followed by academics and non-billionaires, while old influencers show the greatest misalignment.