A 1985 congressional report predicted office jobs would stall

The setup

Congress's own Office of Technology Assessment produced 'Automation of America's Offices, 1985-2000' in December 1985, requested by the Senate Committee on Labor and Human Resources and the House Committee on Education and Labor. The report's foreword frames its subject as roughly 45 million American office workers, and its brief was to assess how continuing, rapid adoption of computers and telecommunications would reshape employment patterns, required skills and the office environment over the following fifteen years. Princeton University's OTA Legacy Archive preserves the report's full text and its original NTIS catalog citation.

What the documents show

The report argues that computers represent a fundamental technological change rather than a marginal improvement, comparing their adoption to the telephone and typewriter rather than the industrial robot, and states that recent employment forecasts from other bodies were flawed and likely understated automation's effects. It argues the potential productivity benefits of office automation are larger than generally recognized, but says those benefits have been and will keep being masked during a transition period by technical and managerial problems and by institutional inertia in adapting to the new technology, a reading the report offers as its own interpretation, not a measured result.

The friction

The report is unusually direct about the limits of its own forecasting tools. It states plainly that existing capability is not good enough to resolve this "very large uncertainty," and warns that recent forecasts "should not lull policymakers into complacency," calling the entire area fraught with uncertainty. It also flags a structural gap: no federal agency, it says, held both the mandate and the resources to combine technological and employment forecasting, leaving the very question the report was asked to answer without an adequate measurement apparatus at the time.

What changed in the work

What the report supports is a documented, contemporaneous baseline: in December 1985, Congress's own technology assessment office expected office automation to be pervasive, potentially productivity-enhancing in ways not yet visible in the data, and disruptive to established occupational categories, while explicitly declining to certify any of those expectations as confident predictions. It does not support treating those forecasts as validated by what actually happened between 1985 and 2000; the report itself never claims that, and this entry does not extend the claim on its behalf. The hedging language reads, four decades later, as a direct precedent for how today's AI-adoption surveys typically qualify their own numbers.

• Does a cited automation-adoption statistic distinguish a measured outcome from a forecast that has not yet been checked against results?
• What specific data gap is the source acknowledging, and has it been closed since, or only restated?
• Does a productivity claim made today distinguish a measured gain from one the report itself would call masked by transition costs and inertia?

Reading a 1985 forecast against the years that followed is tempting, but this report's own text asks for something narrower: notice what it admits it cannot yet measure, and hold today's automation claims to the same standard of disclosed uncertainty.