Companies embracing AI productivity gains struggle with poor leadership guidance and unsustainable token-pricing costs, triggering a shift away from incentivizing AI use.

BCG surveyed nearly 12,000 frontline employees and found 42% reported saving eight hours per week through regular AI use, but 66% received limited to no guidance on how to deploy that saved time, and half said they're not using it for more strategic work.

Companies including Amazon, Meta, Microsoft, and Uber faced surging AI costs tied to token-based pricing models (where fees scale with AI resource consumption). Amazon employees were 'tokenmaxxing'—using AI excessively to meet company metrics—until Amazon scrapped the internal tracking after AI bots were deployed on useless tasks.

BCG's David Martin and other leaders attribute the productivity paradox to senior leaders failing to articulate a clear vision and strategy for AI adoption, which increases employee fear and inhibits sharing of AI techniques across teams. Success requires comprehensive upskilling training and treating AI agents as 'digital employees' rather than tools to reduce displacement anxiety.