The Life and Death of Direct File [pdf]

AI was a side thread rather than the substance of the discussion. Most commenters argued over Direct File’s economics, political cancellation, and whether government ought to build public digital services at all; those claims do not establish much about AI itself. The clearest AI-specific dispute emerged when one participant used an unspecified automated analysis to claim the service cost far more per filing than private preparation. Others challenged the comparison as misleading because the program was a deliberately limited pilot, and one reader specifically faulted the AI-backed claim for omitting both the model and the underlying inputs. The shared lesson was not that an AI calculation is worthless, but that it cannot substitute for transparent assumptions, especially when evaluating a small rollout that has not reached scale. Several readers saw more constructive uses for current models: turning a long post-mortem into an accessible summary and extracting lessons for another workplace. A more ambitious suggestion was that AI-assisted document review and fast prototypes could reduce the decision bottlenecks that consumed much of the project’s schedule. That remained speculative, however. Nobody showed evidence that AI can overcome the lobbying pressure, institutional incentives, and leadership churn that commenters saw as the larger reasons the service disappeared.