When Machines Govern: Why Silicon Valley's Science Fiction Obsession Gets the Future Wrong
8/10/2026
Science fiction has always had a complicated relationship with technology. It warns. It inspires. It satirizes. But it rarely — if ever — hands anyone a blueprint. That distinction matters enormously right now, as a growing number of tech leaders invoke sci-fi visions to justify sweeping changes to the way governments, institutions, and societies operate.
Historian Jill Lepore, a staff writer at The New Yorker and Harvard professor of American history, has stepped into this debate with a pointed critique: Silicon Valley is not just reading science fiction badly — it is using that misreading to erode democratic norms in ways that carry serious long-term consequences.
The "Government by Machines" Problem
At the heart of Lepore's argument is a concept she calls "government by machines" — the idea that algorithmic systems, AI models, and automated decision-making tools are increasingly being positioned as superior alternatives to messy, slow human governance. The pitch is seductive: machines don't get tired, don't take bribes, and don't have bad days. Why not let them run things?
The problem, Lepore argues, is that this framing strips away what governance actually is. Democratic systems are not just mechanisms for reaching efficient decisions. They are processes for surfacing disagreement, protecting minority voices, building consensus, and assigning accountability. An algorithm can optimize for a defined objective — but who defines that objective? That question is inherently political, and automating the answer doesn't make it less so. It just obscures who made the choice.
This is a critique that technologists and robotics professionals would do well to internalize. As AI hardware becomes more capable — from edge inference platforms processing vision and language models entirely on-device, like the NVIDIA Jetson Orin Nano Super, to fully autonomous systems navigating complex physical environments — the temptation to hand over more and more decisions to machines will only grow. The engineering capability is real. The governance question is whether it should be done, and under what conditions.
The Bad Science Fiction Reader
Lepore's specific jab at Elon Musk — that he is a "bad science fiction reader" — is more substantive than it sounds. The canon of science fiction that has influenced Silicon Valley's imagination (Isaac Asimov, Robert Heinlein, Arthur C. Clarke) is routinely read as a kind of aspirational roadmap: here is the future we should build. But Lepore, as a historian, reads these works very differently.
Science fiction, particularly the foundational works of the 20th century, is largely cautionary. Asimov's robot stories are not celebrations of autonomous machines — they are systematic explorations of how rules and systems fail in edge cases, how the letter of a law can be turned against its spirit. The robot that follows its programming perfectly but causes harm anyway is a recurring theme precisely because it is a warning.
When tech leaders cherry-pick the inspiring imagery of science fiction while ignoring its embedded critique, they aren't being visionary. They are, Lepore suggests, being selective readers — and the consequences of that selectivity are playing out in real time across social media algorithms, automated content moderation, AI-assisted hiring, and proposals for algorithmically managed public services.
What This Means for the Robotics and AI Industry
The robotics sector is not immune to these dynamics. Autonomous systems are entering more and more domains: delivery robots navigate restaurant floors, quadrupeds conduct industrial inspections, and drones survey vast agricultural landscapes with minimal human intervention. Each of these deployments raises accountability questions that purely technical specifications cannot answer.
Who is responsible when an autonomous delivery robot causes an incident? How should inspection data gathered by a drone be governed and shared? When an AI-driven agricultural system makes a planting recommendation that fails an entire crop, who bears the cost?
These are not hypothetical edge cases — they are the kinds of questions that need regulatory frameworks, democratic deliberation, and human accountability structures. The robot or drone itself is not the problem. The problem is the assumption, sometimes implicit in how these systems are sold and deployed, that automation replaces the need for human judgment rather than supporting it.
Lepore's intervention is a reminder that the people building and deploying frontier technology have an obligation to engage seriously with governance questions — not to hand those questions off to the technology itself.
The Constructive Path Forward
None of this means autonomous systems are bad, or that robotics innovation should slow down. The practical benefits are too significant to dismiss: drones covering dozens of acres per hour to improve crop yields, quadruped robots handling dangerous inspections so humans don't have to, and educational robots helping the next generation develop the computational thinking skills to engage with an AI-shaped world.
But beneficial technology and accountable governance are not in tension — they require each other. The strongest, most durable deployments of autonomous systems will be the ones built with clear accountability chains, transparent data practices, and genuine public understanding of what the machines are and are not deciding.
That is the science fiction lesson Silicon Valley keeps missing: the stories were never about the machines. They were always about the people who built them, and the societies that had to live with the consequences.
Interested in exploring AI hardware or autonomous systems for your research or operations? Get in touch with our team to discuss what the right platform looks like for your use case.
References
This article was drafted with AI assistance and reviewed before publishing.