Most Americans don’t fear an AI apocalypse, according to new research

An online study examining attitudes towards AI indicated that most people tend to hold optimistic views toward AI and tend to disagree with extreme negative attitudes. Moreover, people with higher social health, higher agreeableness, lower neuroticism and loneliness, as well as those more familiar with technology, tended to have more favorable views about the large-scale impact of AI. The paper was published in the Journal of Technology in Behavioral Science.

As the rapid development of artificial intelligence systems (AI) is visibly transforming how people work and lead their lives, extensive debates are ongoing about what the large-scale impact of AI technologies on the human culture and society will be. Researchers and policymakers debate whether AI will primarily bring major benefits or create serious risks.

One concept in these discussions is P(Doom) – an expectation that advanced AI could cause human extinction or an irreversible collapse of civilization. Different experts assign very different probabilities to P(Doom), ranging from extremely low to relatively high. Some researchers argue that if AI systems become more intelligent than humans and are not properly aligned with human goals, they could create catastrophic outcomes. Others believe that such scenarios are unlikely and that AI will remain under human control through regulation, engineering safeguards, and institutional oversight.

AI optimism is the view that artificial intelligence will largely benefit humanity by accelerating scientific discovery, improving healthcare, increasing productivity, and solving global problems. Optimists point to the historical pattern in which new technologies ultimately increased human prosperity despite temporary disruptions. At the same time, even optimistic perspectives usually acknowledge that AI may bring challenges such as job displacement, misinformation, or concentration of power.

Study authors Rose E. Guingrich and Michael S. A. Graziano wanted to explore the prevalence of P(Doom) concerns in the USA and who holds them. They wanted to investigate how people perceive AI’s impact on themselves and on society, and how are individual psychological characteristics associated with these views.

Free daily newsletter

Study participants were 402 U.S. residents recruited through Prolific in June 2023. 49% of the participants were women. Their age ranged between 18 and 65 years, with most being between 25 and 44. Study authors paid each of them $12 for their participation.

Study participants were randomly divided into two groups. One group was assigned to interact with one of the three popular chatbots (ChatGPT, Replika or Anima) for at least 10 minutes immediately prior to answering the survey questions. Participants from the other group directly proceeded to the survey.

The study survey contained assessments of AI sentiments and p(doom) views (a scale created by study authors), interest and experience with new technologies (the Affinity for Technology Interaction Scale), personality (the Ten Item Personality Inventory), sociability (the Self-Perception Profile for Adults), social competence (the Perceived Social Competence Scale), self-esteem (the Rosenberg Self-Esteem Scale), and loneliness (the UCLA Loneliness Scale). The survey also asked whether the participant had any mental health diagnosis, about demographic data, and participants’ interest in engaging with a chatbot or a human (“I am interested in talking to a chatbot right now” and “I am interested in talking to another person right now”).

Results showed that the participants’ answers tended to be divided. However, the majority of study participants disagreed with p(doom) statements like “I worry that AI is very bad” or “I worry that AI will take over the world”.

Participants also tended to disagree with the statement that “AI agents like chatbots, digital voice assistants, or robots would make good social companions.”, although the ratings were much more evenly distributed. Participants were also divided about whether AI should have moral rights. Study participants generally tended to show somewhat higher agreement with statements describing positive views of the AI.

Further analyses revealed that individuals with higher reported social health (i.e., social competence, sociability, and self-esteem) and agreeableness tended to be more positive about AI. The same was the case with individuals more familiar with technology. On the other hand, individuals with more pronounced neuroticism and loneliness tended to have less favorable views toward the large-scale impact of AI.

“Our research suggests that at this moment in time, in a representative online US sample, extreme, negative p(doom) attitudes are not the norm. The public is generally positive toward AI, in terms of its impact on their personal lives and on society,” the study authors concluded.

The study sheds light on current attitudes people in the U.S. have about AI. However, as the field of AI continues to evolve rapidly, these views may change as a reflection of new experiences.

The paper, “P(doom) Versus AI Optimism: Attitudes Toward Artificial Intelligence and the Factors That Shape Them,” was authored by Rose E. Guingrich and Michael S. A. Graziano.