Science fiction narratives play an important role in shaping how we imagine the future of artificial intelligence, making dark scenarios easier to envision and describe in detail than desirable alternatives, according to a new study by Corvinus University.

Lilla Vicsek and Róbert Pintér, researchers at Corvinus University of Budapest, identified a striking imbalance in how people imagine the future of AI. Between 2021 and 2025, they interviewed 60 university students, and their findings were published in the journal AI & Society in August. Participants were asked to imagine positive and negative futures separately, before considering which scenarios they believed were genuinely likely to occur. 

The differences were striking. Dystopian futures were more detailed, coherent and narrative-driven, featuring scenarios such as machines rebelling against humans, people becoming obsolete, robot dictatorships and social alienation. Around one in three participants referred to a well-known science fiction work to illustrate their vision, mentioning films such as The Matrix, The Terminator and Blade Runner. More than half of those interviewed also felt that imagining a negative future was easier. While negative scenarios sometimes extended 100 or even 200 years into the future, positive visions tended to remain much closer to the present. 

Imagining an idyllic society powered by AI, however, proved considerably more difficult. Positive scenarios were less vivid, often limited to a handful of technological benefits. These included better healthcare, more convenient everyday lives, robots performing dangerous jobs and smarter homes. According to the researchers, these visions mostly involved improved versions of existing technologies rather than fundamentally different societies. Participants also drew on science fiction for positive examples, but generally referred to isolated elements of particular works, including Star Wars, Elysium and Her. 

“This imbalance in our ability to imagine the future may also have wider social implications. If more credible and well-developed positive visions circulated in popular culture, public debate could shift from simply asking ‘What are we afraid of?’ to ‘What kind of future do we want?'” said Lilla Vicsek, Professor at Corvinus University and lead author of the study. 

She added: “At the same time, imagining positive futures with artificial intelligence must not become a means of legitimising existing power imbalances or obscuring the objectives of technology companies. We should aim to ensure that marginalised groups and competing interests have a genuine say in shaping these futures.” 

Science fiction provided participants with a framework for thinking and talking about a technological future that is difficult to predict. However, the emergence of generative AI has partly changed this. Those interviewed in 2024–2025 were more likely to draw on their own experiences and relied less exclusively on science fiction narratives to shape their expectations. 

For example, earlier participants often described artistic creativity as a distinctly human domain that machines were unlikely to master. In later interviews, this confidence had largely disappeared, replaced by concerns that generative AI was already entering creative fields previously considered uniquely human. 

As the study was based on a small, non-representative sample, its findings cannot be statistically generalised to society as a whole.