[ACC] Will Automation Lead To Economic Crisis?

A 2019 Adversarial Collaboration Contest entry by Doug Summers-Stay and Erusian on whether automation/AI will cause a general sustained economic crisis (Erusian defending 'no,' Doug challenging). The tldr framing is the useful core: until the pace of automation outruns new-job creation there's no reason to expect mass unemployment; the economy only truly breaks when AI can learn any new job as fast and cheaply as a person -- but at that point everything breaks, because that's the Singularity. The survey is comprehensive and fair: current technological unemployment (employment rates are historically normal; three prior automation waves -- agriculture 70%->few %, manufacturing 40%->10%, housework 60->15 hrs/week); the recurring 'computers will never do X' claims paired one-for-one with research papers attempting each X (a motte-and-bailey where the defensible motte is phenomenal consciousness, the bailey is discriminating/simulating emotion, which neural nets can do); which jobs go first (Frey's 49% vs Arntz's 9%); the economics (Say's law, the 'theory of compensation,' Marx's objection, capital owners capturing productivity gains, the post-1975 wage-vs-productivity divergence); jobs people may pay a premium for a human to do; and the post-Singularity endgame where economic models break down. Solid, balanced, and informative -- but more conventional/survey-like than the standout ACC (SSC-832), reaching fairly standard conclusions (compensation theory holds until the Singularity) without an original model.

Quality 71 · Strong. Strong (71): a thorough, balanced, well-organized adversarial collaboration that genuinely clarifies the automation-economics debate, with a useful framing of the crisis boundary (no crisis until AI matches human learning speed/cost = Singularity) and a sharp motte-and-bailey diagnosis of impossibility claims. Held in the Strong band because it surveys and synthesizes fairly standard economics (compensation theory, Say's law) and lands on conventional conclusions, without the quantified model or genuine both-sides-updating of the standout ACC (SSC-832, 78).

Claude’s paradigm shift 46 · Moderate. Moderate (46): the careful synthesis and the 'crisis only at the point AI learns jobs as fast as humans' boundary are a fresh, clarifying framing, but the underlying economics (compensation theory, Say's law, Frey/Arntz) is conveyed rather than novel.

Real-world impact 2 · Minor. A thorough, balanced adversarial collaboration that clarifies the automation-economics debate with a useful framing (no general crisis until AI matches human learning speed/cost — which is the Singularity) and a motte-and-bailey diagnosis of impossibility claims. Conceptual influence within economics discourse, no material change — low RWI.