Uber kept its AI spending "broadly stable" in the second quarter, CFO Balaji Krishnamurthy said on Wednesday, by "setting better defaults for different use cases," using lower-cost models for some tasks, and "letting employees more clearly understand and manage their spend." As a result, "cost per token has declined over the past several months, even as adoption has continued to increase." CEO Dara Khosrowshahi said the company is using AI for "thousands of small hits and improvements" rather than a "giant hit." One example is destination suggestions when customers open the app, which correctly predicts the ride's destination three-quarters of the time. Uber laid off about 10% of its customer service workers last month, citing efficiency gains from AI, and Krishnamurthy called customer service "a clear area where we should be able to up the quality of our customer support interactions as well as reduce the cost of our effort." The company has also used "agentic pods" to find AI uses from finance to HR. Uber facilitates "hundreds of thousands" of rides in self-driving cars each week, less than 0.5% of the 300 million trips through the app in the same period, Khosrowshahi said, noting robotaxis require real-world testing and government approvals.