At the recent World Economic Forum held in Davos, Switzerland, Dara Khosrowshahi, CEO of Uber Technologies Inc., presented a candid assessment of how organizations are engaging with artificial intelligence. Observing that a pronounced divide has emerged between companies merely discussing AI and those genuinely altering their operations with the technology, Khosrowshahi characterized many efforts as essentially "play-acting."
He noted that while the majority of businesses are increasing their financial commitments toward AI, many continue to operate their legacy processes with AI appended superficially rather than fundamentally rethinking workflows. This disparity hinges on whether a company incorporates AI into existing systems or chooses to overhaul its methods to built around AI capabilities.
In conversation with Bloomberg’s Head of Economics and Government, Stephanie Flanders, Khosrowshahi remarked that the dominant AI initiatives often involve "the easy stuff" – for example, leveraging AI tools to summarize client presentations or automate routine, low-complexity tasks. Though such applications can yield incremental improvements, they fall short of unlocking AI’s full transformative potential.
The real challenge, he stressed, lies in reengineering processes from the ground up so that AI is integrally embedded, enabling more advanced reasoning and adaptability.
Data from a Boston Consulting Group report underscored this momentum: companies, on average, plan to more than double their AI-related expenditure this year, with AI investment rising from roughly 0.8% to 1.7% of revenue. Notably, approximately 90% of organizations intend to allocate additional funds toward AI despite encountering difficulties in measuring returns. This marks a transition away from experimental deployments toward more widespread integration across enterprises.
Reflecting on Uber’s own journey, Khosrowshahi revealed that significant progress only emerged after abandoning early AI strategies. Initially, the company employed AI systems designed to follow established customer service policies. While these efforts brought some gains, they did not realize the substantial improvements desired.
“Allowing the AI actually to reason through that and throwing away all of the old policies is turning out to be the most promising way forward,” he explained. Uber rebuilt its customer service frameworks by replacing inflexible rules with objective-focused AI agents, specifically aiming to enhance the customer’s emotional experience following each interaction.
Khosrowshahi further elaborated on the broader implications: from a structural perspective, a company is essentially a collection of policies and rules. Achieving the true capabilities AI offers requires breaking away from these precepts and starting anew.
Within Uber, developers employ advanced AI tools such as Anysphere’s Cursor, an AI-driven coding assistant, and Anthropic’s Claude, a sophisticated large language model, to assist in this transformation.
However, the path to AI integration is not without internal turbulence. The CEO metaphorically referred to the process as surviving “a bunch of car crashes internally.” Indeed, while workers often value AI’s utility, surveys like KPMG's American Worker Survey reveal growing apprehension about job security and skills erosion due to AI’s impact.
In summary, Uber’s experience illustrates that simply adding AI on top of existing systems is insufficient. Far-reaching changes necessitate breaking current rules and reinventing procedures with AI deeply embedded, a complex and challenging transformation journey facing many companies as investment in AI technologies accelerates globally.