{"slug": "experts-and-the-public-have-radically-different-visions-of-an-automated-future", "title": "Experts and the public have radically different visions of an automated future", "summary": "A new survey published in AI & SOCIETY reveals that AI experts are significantly more optimistic about the future of automation than the general public, with experts perceiving lower risks and greater benefits. The study, led by Philipp Brauner of RWTH Aachen University, surveyed 1,110 German citizens and 119 AI researchers, finding systematic differences in how each group evaluates potential AI scenarios over the next decade.", "body_md": "Artificial intelligence is transforming communities, but the people who study these systems and the people who use them harbor profoundly different expectations about the future. A recent survey reveals that technical experts generally possess a more optimistic vision of artificial intelligence than everyday citizens, prioritizing potential benefits over associated hazards. The findings were published in the journal [ AI & SOCIETY](https://doi.org/10.1007/s00146-026-03023-8).\n\nLead author Philipp Brauner, a communication researcher at RWTH Aachen University in Germany, and his colleagues wanted to understand how varying groups imagine the trajectory of automation. The researchers focused on mapping the mental models of different populations. A mental model is a person’s internal understanding of how something works, what capabilities it possesses, and what consequences it might bring to reality. By comparing these internal frameworks, the team hoped to uncover potential friction points in technology adoption.\n\nWhen engineers design tools based primarily on their own optimistic mental models, they run a specific risk. They might unwittingly build what the researchers describe as procrustean artificial intelligence. The term references the Greek myth of Procrustes, a rogue blacksmith who forced his guests to fit an iron bed by stretching their limbs or amputating them. In a modern technological context, it describes systems that force the public to adapt to rigid technical constraints rather than bending the actual technology to fit diverse human needs.\n\nTo map these underlying views, the research team recruited 1,110 members of the German public alongside 119 academic artificial intelligence experts. The public group represented a cross section of society in terms of age and socioeconomic status. The expert group consisted mainly of researchers from Germany and the Netherlands who actively inform policy, educate practitioners, and build algorithms.\n\nEach participant completed an online survey evaluating a random selection of statements from a master list of 71 hypothetical scenarios. To build this comprehensive list, the researchers drew upon sociological frameworks that divide society into core subsystems. These subsystems included the economy, the legal system, science, politics, religion, and education. Ensuring every sector was represented, the finalized scenarios covered a wide array of potential applications and impacts expected over the next decade.\n\nTopics ranged from routine automation, like algorithms driving cars and tutoring students, to more speculative outcomes. Some participants evaluated the likelihood of software acting in accordance with moral concepts or solving social issues. Other scenarios presented dystopian possibilities, such as machines destroying humanity, increasing personal loneliness, or making independent decisions about human life and death.\n\nParticipants evaluated each assigned scenario across four distinct dimensions. They estimated the likelihood of the event occurring within ten years. They also evaluated the personal and societal risks involved, the potential benefits, and their overall positive or negative feeling about the outcome.\n\nThe results highlight systematic differences between how specialists and laypeople evaluate technological progress. On average, the experts rated the 71 scenarios as more likely to occur than the public did. At the same time, the technical specialists perceived lower personal risks and anticipated greater benefits across the board. This translated into an overall more positive evaluation of artificial intelligence among the researchers compared to the everyday users.\n\n[Add PsyPost to your preferred sources](https://www.google.com/preferences/source?q=psypost.org)\n\nExperts also displayed a much wider variety of opinions depending on the specific scenario at hand. The specialist evaluations varied wildly from highly positive to highly negative, suggesting a highly differentiated view of the technology. The public, by contrast, rated nearly everything closer to a baseline of generalized concern.\n\nBeyond scoring things differently, the two groups used different internal formulas to weigh the good and the bad. Both populations exhibited what psychologists call the affect heuristic. This is a mental shortcut where individuals view highly beneficial things as inherently less risky, and highly risky things as less beneficial. Because of this psychological mechanism, risk and benefit scores were inversely related in both groups.\n\nWhen deciding the overall value of a specific scenario, experts were heavily influenced by the perceived advantages. In the specialist group, the anticipated benefits drove their final judgments about three times as strongly as the perceived risks did. A high expectation of risk barely moved the needle on an expert’s overall enthusiasm for a tool, provided the associated benefits were also high.\n\nFor the general public, the calculation looked quite different. Everyday citizens also prized utility and convenience, but they were far more sensitive to potential downsides. In the public sample, perceived risks exerted a much heavier drag on the overall value of a scenario. When everyday users sensed danger in a technological application, that threat heavily dampened their general approval.\n\nThis divergence in risk calculation led to notable disagreements on specific topics. Citizens expressed high expectations and deep concern regarding existential threats. They feared applications that might replace human relationships, control information, or lead to a total loss of human control. Conversely, experts considered these apocalyptic outcomes highly unlikely and rated them poorly.\n\nInstead, the academic group showed sweeping optimism about structural and scientific transformations. The specialists assigned high probabilities and high value to scenarios where artificial intelligence improves healthcare, boosts environmental sustainability, and supports medical decision making. They viewed these positive outcomes as both highly probable and immensely useful.\n\nDespite these contrasting visions, the two groups did agree on several specific issues. Both samples recognized severe dangers in the potential for criminals to misuse the technology. They also shared a positive view of artificial intelligence carrying out medical diagnoses. The researchers suggest that identifying these narrow areas of consensus can help policymakers focus their regulatory attention where it is needed most.\n\nIn an exploratory analysis, the research team grouped the 71 scenarios based on how participants rated them, searching for broader trends. They found two main clusters of technological applications. The first cluster included supportive systems, such as health monitors and smart city planners, which both groups generally approved of. The second cluster featured autonomous, dominant systems capable of making independent political decisions or determining warfare. While both groups recognized these clusters, the public was much faster to assign the autonomous scenarios into an overarching category of severe threat.\n\nLike any survey based on hypothetical futures, the study has limitations. The participants evaluated imaginary scenarios in passing, meaning the results capture immediate emotional reactions rather than rational forecasting. These emotional heuristics still strongly dictate how people adopt and trust new tools, but they do not reflect guaranteed outcomes. Evaluating highly abstract vignettes also removes the nuance of how a tool is actually built.\n\nAdditionally, the survey primarily sampled participants from Germany in early 2023. The data collection occurred shortly after generative language models sparked a massive surge in global media coverage. Public sentiment often shifts rapidly alongside such intense news cycles, meaning these specific technological anxieties might have evolved since the data was gathered.\n\nFuture investigations will need to track these opinions over time and across different cultures. Expanding the participant pool to include non-Western countries could reveal whether these risk profiles represent universal human anxieties or specific regional attitudes. The current researchers also advocate for incorporating qualitative interviews to better understand why everyday users group certain threats together.\n\nUltimately, the research points to a pressing need for participatory design practices in software development. If future algorithms are built solely according to the optimistic visions of technical specialists, they might inadvertently neglect the valid safety concerns of the people who actually have to use them. Bridging this perception gap could help modern society integrate automation in a way that truly respects human priorities.\n\nThe study, “[Charting the AI perception gap: divergent views on risk, benefit, and value between experts and the public challenge the societal acceptance of AI](https://doi.org/10.1007/s00146-026-03023-8),” was authored by Philipp Brauner, Felix Glawe, Gian Luca Liehner, Luisa Vervier, and Martina Ziefle.", "url": "https://wpnews.pro/news/experts-and-the-public-have-radically-different-visions-of-an-automated-future", "canonical_source": "https://www.psypost.org/experts-and-the-public-have-radically-different-visions-of-an-automated-future/", "published_at": "2026-07-01 13:00:27+00:00", "updated_at": "2026-07-01 13:32:46.655527+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-safety", "ai-ethics", "ai-policy", "ai-research"], "entities": ["Philipp Brauner", "RWTH Aachen University", "Germany", "Netherlands", "AI & SOCIETY"], "alternates": {"html": "https://wpnews.pro/news/experts-and-the-public-have-radically-different-visions-of-an-automated-future", "markdown": "https://wpnews.pro/news/experts-and-the-public-have-radically-different-visions-of-an-automated-future.md", "text": "https://wpnews.pro/news/experts-and-the-public-have-radically-different-visions-of-an-automated-future.txt", "jsonld": "https://wpnews.pro/news/experts-and-the-public-have-radically-different-visions-of-an-automated-future.jsonld"}}