Fantasy Inventions
Fantasy inventions are the imagined artifacts that precede the real ones by decades or centuries — Daedalus's wings, Leonardo's ornithopter, Jules Verne's submarine, H.G. Wells's atomic bomb, Edward Bellamy's credit card, Mary Shelley's Frankenstein, and the long lineage of thinking machines from the golem through Hal 9000 to Samantha from Her. Basalla argued that these imagined artifacts are not merely entertainment. They constitute a reservoir of potential variation that operates through two channels: they shape the direction of inventive effort by marking certain artifacts as desirable, and they prime the selection environment by establishing cultural expectations that the real artifact can eventually fill. The fantasy does not predict the future. It shapes the niche that the future artifact must fill.
In the AI Story
The directional channel operates because inventors do not explore the space of possible artifacts randomly. They explore in directions that their culture has already marked as desirable, and the marking is done in large part by the accumulated weight of fantasy inventions. The Wright brothers did not set out to solve an abstract problem in aerodynamics. They set out to build the thing their culture had been imagining for centuries: a machine that flies. The image preceded the artifact. The fantasy constrained the direction of variation by establishing what a culture believed technology should eventually be able to do.
The receptive channel operates because a society that has been imagining an artifact for decades or centuries is predisposed to adopt it when it finally arrives. The selection environment has been primed. When the airplane appeared in 1903, the public response was not bewilderment but recognition. The adoption was accelerated by the prior existence of the fantasy, which had created a cultural niche the real artifact could fill without requiring the society to first develop the conceptual framework for understanding what it was.
The AI case illustrates both channels with exceptional clarity. The thinking machine is among the oldest fantasy inventions in human culture — the golem, Frankenstein, Čapek's robots, Hal 9000, Data, Samantha, the replicants of Blade Runner. When ChatGPT reached fifty million users in two months, the fastest adoption of any technology in history, it was entering a selection environment prepared for decades by these fantasy inventions. The public did not need to be taught what a conversational AI was. They had seen it in a hundred films, read it in a thousand novels. The cultural niche already existed.
But the fantasy shapes more than adoption speed. It shapes the kind of AI the selection environment favors. The dominant fantasy of artificial intelligence is the autonomous mind — an intelligence that thinks independently, that has its own goals, that may or may not be aligned with human interests. This fantasy shapes user expectations in ways that are largely invisible to the people operating within it. Users anthropomorphize AI systems, attributing intentions, preferences, even feelings to mathematical processes that have none of these things. They evaluate systems by the criteria fantasy has established: Does it feel like talking to a mind? Does it seem to understand me? These narrative criteria confer adoption advantages on systems that satisfy them, regardless of whether such systems are the most beneficial tools for the tasks at hand.
The practical implication is uncomfortable for the technology industry, which tends to treat science fiction as a source of product ideas rather than as a shaping force on the selection environment. The science fiction of the last century gave us the autonomous-mind fantasy almost exclusively. It did not give us — or gave us only in scattered, peripheral works — the fantasy of the collaborative tool, the amplifier of human judgment, the instrument that makes the practitioner more capable rather than replacing the practitioner entirely. The fantasy inventory is lopsided. The selection environment it has shaped is correspondingly lopsided, primed to receive AI that looks like an autonomous mind and less prepared for AI that operates as an extension of human cognition.
Origin
Basalla treated fantasy inventions most fully in chapter 6 of The Evolution of Technology. The concept draws on the history of technology transfer between speculative thought and engineering practice, and on the broader cultural-studies tradition that takes science fiction seriously as a shaping force on material culture.
Key Ideas
Fantasy inventions precede real ones. The imagined artifact often exists in culture for centuries before the technical means to build it arrive.
Fantasies shape direction. The imagined artifacts a culture accumulates mark the directions its inventors will attempt to explore.
Fantasies prime the selection environment. The society that has been imagining an artifact is predisposed to adopt it when it arrives, because the cultural niche already exists.
The fantasy inventory for AI is lopsided. A century of science fiction has given us almost exclusively the autonomous-mind fantasy, and almost nothing about AI as collaborative tool.
Reshaping the inventory matters. The stories a culture tells about its technological future constrain the future it is capable of building.
Appears in the Orange Pill Cycle
Further reading
- George Basalla, The Evolution of Technology, chapter 6 (Cambridge University Press, 1988)
- David Nye, American Technological Sublime (MIT Press, 1994)
- Howard Segal, Technological Utopianism in American Culture (University of Chicago Press, 1985)
- Sherry Turkle, Alone Together (Basic Books, 2011)