Australia’s biggest bank says corporate AI is racking up bigger bills and producing ‘work slop’
The business move
Matt Comyn, chief executive of the Commonwealth Bank of Australia (CBA), flagged rising pain points in corporate AI adoption during a recent speech. He described the quality of AI-generated work flowing through large enterprise workflows as “work slop.” At the same time, CBA is seeing AI usage costs escalate sharply because token-based billing scales with task complexity. This reflects growing challenges for big companies deploying AI at scale in real-world processes.
Why it matters
The phrase “work slop” captures a practical problem: AI outputs are often low quality, requiring human rework or correction. This degrades the efficiency gains AI promised. Meanwhile, token billing for AI models ties costs directly to input and output size and complexity. As businesses push AI to handle more complex, nuanced corporate tasks, their AI expenses balloon. This squeezes budgets and creates tension between the drive to automate and the need to keep costs manageable. CBA’s experience signals a broader reckoning with AI economics and output quality in enterprise use.
Who gains and who gets squeezed
Vendors may benefit in the short term from rising token usage, but if output quality remains poor, buyers will push back or limit AI integration. Banks, insurers, and other complex service enterprises face particular risk getting stuck with high bills for underwhelming AI results. AI providers able to improve fine-tuning, specialize models for specific domains, or offer predictable pricing could gain an edge. Buyers who build human-in-the-loop safeguards or careful AI task scoping will manage cost and quality risks better.
What to watch next
Operators will watch how the economics of corporate AI evolve as token costs rise with workload complexity. Expect increasing emphasis on tuning AI to reduce low-value outputs and refining workflows for more efficient token use. Pricing innovations or contracts that limit runaway costs could become a priority for enterprise buyers. Also, enterprise buyers will push vendors harder on demonstrable improvements in output quality to stop “work slop” from eroding trust and productivity.
AI Quick Briefs Editorial Desk
