TLDR: A two-wave survey study in Switzerland found a significant decrease in public acceptance of AI and an increased demand for human oversight after the generative AI boom (post-ChatGPT launch). The study, involving a representative Swiss sample, revealed that the proportion of respondents finding AI “not acceptable at all” rose from 23% to 30%, and support for human-only decision-making increased from 18% to 26%. These shifts were particularly pronounced in high-impact scenarios and amplified existing social inequalities across educational, linguistic, and gender lines, challenging industry assumptions about public readiness for AI deployment.
The rapid integration of generative artificial intelligence (GenAI) into various products and services has been a defining trend in recent years. However, a new study reveals a surprising consequence: a significant decrease in public acceptance of AI and a growing demand for human oversight, especially in critical decision-making scenarios.
Researchers Joachim Baumann, Aleksandra Urman, Ulrich Leicht-Deobald, Zachary J. Roman, Anikó Hannák, and Markus Christen conducted a large-scale two-wave survey in Switzerland, comparing public attitudes before (January-February 2022) and after (July-August 2023) the launch of ChatGPT and the subsequent GenAI boom. Their findings challenge the assumption that the public is readily embracing widespread AI deployment.
A Decline in AI Acceptance
The study found a notable drop in overall AI acceptance. The proportion of respondents who considered AI “not acceptable at all” increased from 23% in the first survey wave to 30% in the second. Conversely, those who found AI “definitely acceptable” decreased from 13% to 10%. This shift was particularly evident in high-impact decisions such as prisoner release, therapy discontinuation, and hiring, which consistently showed the lowest acceptance levels.
Even in scenarios with higher initial acceptance, like fake news detection and insurance premium calculations, there was a pronounced increase in people finding AI use unacceptable. This suggests a growing skepticism across various applications, driven by heightened concerns about the reliability and ethical implications of AI technologies, especially in sensitive contexts.
Increased Demand for Human Oversight
Accompanying the reduced acceptance is a significant increase in the public’s demand for human oversight in AI decisions. While human-AI collaboration (where humans are either ‘in the loop’ or ‘on the loop’) remains the preferred model for many, the percentage of people advocating for “human-only” decision-making rose from 18% to 26% after the GenAI boom.
In high-stakes areas like medical diagnosis, therapy discontinuation, hiring, and prisoner release, approximately 40% of respondents in the second wave wanted only humans to be involved. This trend underscores a clear public preference for human accountability and intervention, indicating a desire for humans to retain primary control, particularly when decisions have profound consequences.
Amplified Social Inequalities
The study also highlighted that the GenAI boom has exacerbated existing social inequalities in AI acceptance, widening educational, linguistic, and gender gaps.
- Education Gap: Individuals with higher education levels consistently showed greater acceptance of AI. However, this gap widened post-boom, with lower-educated individuals becoming more critical of AI use. For instance, in medical diagnosis, acceptance among those with general education dropped significantly, while university-educated respondents maintained relatively stable views.
- Language Region Gap: French-speaking respondents demonstrated lower AI acceptance compared to German speakers, a difference that became more pronounced in medical contexts after the GenAI boom. This suggests that cultural or media framing differences might play a role in shaping perceptions.
- Gender Gap: Women expressed significantly more skepticism toward AI, particularly in health-related scenarios like medical diagnosis and therapy discontinuation. This gender gap deepened after the GenAI boom, with women showing a more substantial decrease in acceptance compared to men.
Also Read:
- Navigating the Nuances of AI Trustworthiness: A Review of Leading Ethics Research
- Unveiling Bias: A Deep Dive into Race and Gender Representation in AI-Generated Occupational Personas
Implications for AI Development and Governance
These findings underscore the critical importance of aligning technological development with evolving public preferences. The study suggests that industry assumptions about public readiness for AI deployment may be flawed. The rapid shift in public sentiment highlights the need for adaptive regulatory frameworks that can keep pace with technological advancements and public opinion.
The researchers emphasize that AI definitions, media coverage, and socio-demographic factors all play a role in shaping public attitudes. They advocate for more granular, context-specific regulatory approaches rather than broad, monolithic AI policies, acknowledging that acceptance varies significantly across different application domains.
For a deeper dive into the methodology and detailed results, you can read the full research paper here: Reduced AI Acceptance After the Generative AI Boom: Evidence From a Two-Wave Survey Study.
