Why AI Tools Fail: The Decision Load Problem

Cognizant estimates AI could unlock $4.5 trillion in US labor productivity. Our argument: tools that ignore cognitive load may leave much of it uncaptured

Content Ready for Growth Claude Implementation

Frontend Infrastructure Complete: This blog post template is ready for Growth Claude to add MKT-1 content.

Template Features Implemented:

  • ✅ Schema.org BlogPosting structured data
  • ✅ Section 35 accessibility compliance (44px targets, focus indicators)
  • ✅ Mobile-responsive design with CTE brand consistency
  • ✅ Research callout components for statistics
  • ✅ SEO optimization for AI productivity keywords
  • ✅ Plausible analytics integration
  • ✅ Social media meta tags ready

Growth Claude can replace this placeholder with MKT-1 content following the established design patterns and research callout components.

$4.5T
US labor productivity AI could unlock with today's capabilities (an estimate of potential, not of losses)
Cognizant, “New Work, New World 2026” (January 2026)

Corrections, 28 September 2026: An earlier version said $4.5 trillion in AI productivity gains “remain unrealized because AI tools optimize output without measuring cognitive load” and credited this to Cognizant, Workday and IBM. Cognizant's report estimates the value of work AI could handle today; it does not attribute any shortfall to cognitive load, and the figure does not come from Workday or IBM. We have corrected the wording and the source.

Experience the Missing Measurement

See how CTE measures the cognitive load your AI tools create

Take the Governance Stress Test Read the Framework

Get notified about new EAA series parts

One email when a new position-paper part or field note publishes. No newsletter, no marketing list.

This tool has been retired

It was part of our original decision-load research. See our current work on agent governance instead.

Take the Governance Stress Test