Every generation produces its own version of the Luddite argument — the instinct to slow technological progress through punitive measures rather than adapt to it. The latest iteration comes in the form of proposals to levy a so-called "compute tax" on the processing power that drives artificial intelligence systems, a policy idea the Reason Foundation has sharply criticized as fundamentally misguided.
The historical parallels are striking. When the printing press disrupted the scribal profession, some European authorities imposed restrictions on presses. When the steam engine threatened traditional craftsmen, Britain's Locomotive Acts forced early automobiles to be preceded by a man waving a red flag. In each case, the regulatory impulse to tax or throttle a transformative technology ultimately delayed benefits without preventing disruption.
Compute — the raw processing capacity that trains and runs AI models — has become the new target for policymakers anxious about automation's impact on employment. The logic seems intuitive: if AI is replacing workers, tax the fuel that powers it. But economists and technologists have long noted that taxing capital inputs tends to push development offshore rather than protect domestic jobs, a dynamic well-documented during earlier waves of automation policy in manufacturing.
The Reason Foundation's critique lands within a broader debate the AI field has been wrestling with since at least the early automation scares of the 1960s, when President Kennedy's administration convened commissions to study the "cybernation" threat to labor. Then, as now, the most durable solutions involved workforce investment and adaptation rather than technological restriction.
What distinguishes the current moment is the speed at which AI capabilities are advancing, compressing timelines that once gave societies decades to adjust. That urgency is real. But history suggests that compute taxes would function less as worker protection and more as an innovation tariff — one that wealthier, less-regulated jurisdictions would eagerly exploit, leaving the workers the policy intended to shield in a worse competitive position than before.
The debate over how societies should govern AI's economic disruptions is only beginning. Taxing the hardware may feel satisfying, but the archive of technological history counsels a more measured approach.