A New Metaphor for AI

I recently visited the VINT conference where Franziska Kohlt, who specializes in the history of science, talked about the AIs of past centuries: automata.¹ She traced the development of automata from 8th century BC Greek mythology through the Renaissance to the Industrial Revolution, showing how these mechanical creations opened new ways of thinking about animals and humans. She discussed how Descartes, looking out his window in Amsterdam, wondered whether the figures he saw beneath cloaks and gowns were real humans or mechanical automata dressed as people.² Later, Thomas Huxley took this further, arguing that animals were automata and that our human consciousness is nothing more than a byproduct of our mechanical machinery.

During the discussion after her talk, she made a point that struck me. When asked about implications for today's AI, she said we need a new metaphor for thinking about our intelligent companions. The dominant one for the past two decades has been the Terminator - Skynet coming back to destroy us all. Metaphors shape our thinking and determine the opportunities we see. She gave another example: when people talk about declining insect populations as an "insect apocalypse," it makes the trend seem inevitable, as if there's nothing humans can do about the demise of all insect life. The metaphor hides the fact that we're responsible for the decline and removes our agency. The same thing happens with the AI discourse when we default to the Terminator metaphor.

So what is a better metaphor for thinking about AI? In the 17th century Newton said, "If I have seen further, it is by standing on the shoulders of giants."³ He was implying that his contributions to science were only possible because he could leverage the work of Galileo and Kepler and others. I'm not sure if this is an accurate metaphor for science, but it's a popular one: scientists build on the works of geniuses that came before them to look further and see new paradigms. It's a collaborative viewpoint of what makes scientific progress possible.

With the arrival of intelligent systems, something new has entered our discourse of scientific progress. AIs are being used as assistants and many think they will impact our human endeavors - not just scientific research, but medicine, arts and technology. Their main contributions may be that they compress enormous amounts of human knowledge into a single interface. In that sense, they are giants - of information and compressed human insight. I've been experimenting with this myself while programming: when I use them, I'm not handing over control, I'm climbing on top of their capabilities. They give me leverage: faster understanding, broader context, quicker iteration. But I'm still the one doing the looking, deciding where to go next.

This is where the metaphor fits: it shows how AIs can take us to new heights while preserving our human agency. It acknowledges that we're now standing on something new, something much taller than ever before, and it raises the question: where do we, as humans, want to go next on the shoulders of our AI companions?