Episode Summary
Executive Summary: Kara Swisher interviews Uber CEO Dara Khosrowshahi about how AI powers Uber today and how autonomous vehicles could reshape mobility, delivery, labor, regulation, and city design. Khosrowshahi argues Uber is already an applied-AI company, sees AVs as safer and eventually cheaper than human driving, and believes Uber’s role is to be a platform connecting demand, partners, and eventually fleets—while managing a long, uneven transition for drivers and cities.
Main Topics: Uber as an applied AI company (Priority: 5/5): Khosrowshahi explains that AI has long powered Uber’s core functions, including pricing, matching, routing, identity verification, search, and Uber Eats personalization. He frames Uber as a real-world technology company using AI in physical operations. AI for developer productivity and internal operations (Priority: 5/5): He says the biggest current AI impact is internal: coding, code review, documentation, and on-call engineering. AI agents now help diagnose system issues, though some use cases initially failed because of hallucinations and duplication of work. Autonomous vehicles as Uber’s biggest long-term opportunity (Priority: 5/5): Khosrowshahi argues AVs will become safer than humans, improve over time, and eventually become cheaper. He sees Uber as a platform that should aggregate both human and robot drivers rather than manufacture cars itself. Partnership strategy versus owning the AV stack (Priority: 4/5): Uber is partnering with multiple AV firms and expects both direct channels and marketplace distribution to coexist. Khosrowshahi says Uber benefits from demand, routing, and fleet utilization without needing to build the cars or own the full stack. Regulation, safety, congestion, and urban access (Priority: 4/5): The conversation focuses on what cities and regulators should do as AVs scale: preserve accessibility, manage congestion, and ensure equitable rollout beyond wealthy districts. Khosrowshahi argues Uber’s density can help extend service to underserved areas. Labor transition and job displacement (Priority: 4/5): Swisher presses on the impact of automation on drivers and workers. Khosrowshahi says Uber should be honest about the transition, slow hiring where AVs are arriving, and create alternative work such as AI labeling and data-related tasks. Uber Eats, delivery robotics, and multimodal automation (Priority: 3/5): Khosrowshahi says delivery may become even larger than mobility, with a mix of sidewalk robots, AVs, and drones for different distance bands. He emphasizes first-mile/last-mile logistics and experimentation rather than one universal solution.
Key Arguments: Uber is already an AI company in practice because its core services depend on machine learning for pricing, matching, routing, fraud detection, and personalization. The most immediate AI value at Uber is productivity: AI tools make engineers more efficient, especially in coding and incident response. Initial AI customer-service automation did not work well because humans distrusted model outputs and duplicated work; Uber is moving toward more direct AI execution. AVs will ultimately be safer than human drivers because they do not get tired, distracted, or impaired, and their software improves continuously. Uber should act as a platform for both human and robot drivers rather than become a car manufacturer or vertically integrate the AV stack. A hybrid network is the best transition model because it preserves service quality, keeps humans employed, and gradually introduces AVs to customers. AV deployment raises social questions beyond technology, especially who gets access first, how congestion is managed, and how cities adapt infrastructure and transit policy. Uber sees opportunity in enabling alternative work for displaced labor, including AI labeling, monitoring, and other flexible tasks on the platform. Uber Eats may become larger than mobility, with AVs, sidewalk robots, and drones each solving different parts of the delivery problem. The EV transition remains important, and AV adoption may accelerate EV usage because autonomous vehicles are electric by default.
Data Points: AI usage among developers: 80–90% - Khosrowshahi says most Uber developers are using AI developer tools. Uber operates in countries: 70 - He cites Uber’s global operating footprint when describing on-call engineering challenges. Uber operates in cities: 15,000 - Used to illustrate the scale of operational complexity Uber’s systems handle. Current global ride volume: 35 million rides per day - Khosrowshahi contrasts current human-driven scale with still-small AV volume. AV rides run rate: millions of rides a year - He says Uber is already doing AV rides at a meaningful but small scale. AV share by 2030 (U.S. estimate): ~10% of trip volume - Khosrowshahi says AVs may reach roughly this level in the U.S. by 2030. AV safety comparison: 10x safer / 50x safer scenario discussed - Swisher raises the threshold for whether humans should still be allowed to drive if AVs are far safer. U.S. annual automobile fatalities: ~35,000 per year - Khosrowshahi uses this as the baseline for comparing AV safety. AV fatality thought experiment: 10 deaths per day if 10x safer - He frames the societal acceptance problem even with large safety gains. Uber drivers on network: over 8 million - He references the scale of human labor on Uber’s platform. Driver count later restated: 9 million+ - He uses a rounded higher estimate later when discussing long-term transition. Uber Eats share of business: almost 50% - Khosrowshahi says Eats has grown from about 10% to nearly half of Uber’s business. Uber Eats early share: 10% - Referenced as the starting point before expansion. North America EV penetration: 9% - He says Uber’s North American fleet is only at 9% EV penetration after earlier targets. Europe EV miles: 15%+ - He says Europe is progressing better on EV adoption than the U.S. Customer-service AI mistake rate: about 5% - He estimates the AI makes mistakes or hallucinates in a small but meaningful share of cases. AI productivity gain estimate: 20–30% - He uses this as the productivity uplift from AI for engineers. AV economics in cities: 99% of Uber drivers surpassed by average AV utilization in Austin/Atlanta - He says average Waymo performance exceeds most human drivers on trips completed. AV transition horizon: 10–15 years for meaningful scale; 20–30 years for eventual dominance - He gives multiple timeframes for AV maturation and market shift. Lyft CEO quote referenced: zero likelihood of replacement anytime soon - Swisher cites a contrasting view from Lyft’s CEO on AV readiness. Tesla crash settlement/award cited: over $240 million - Used to highlight liability and safety stakes in autonomous driving.
Pivotal Quotes: "AI has been part of our genetics for a very, very long time." — Dara Khosrowshahi: He explains Uber’s long-standing reliance on machine learning before the current AI boom. "I think eventually AVs will win." — Dara Khosrowshahi: He states his long-term view that autonomous vehicles will replace human driving over time. "If an AV is provably 50 times safer than a human being, do you think we should allow human beings to drive?" — Kara Swisher: She presses the central safety and policy question framing the whole interview.
Implications: Uber is positioning itself as the demand layer for an AI-driven transportation economy. The likely future is hybrid: human drivers, AVs, delivery robots, and drones coexisting while regulators, cities, and labor markets adapt.
About Pivot
With great power, comes great scrutiny. Every Tuesday and Friday, journalist Kara Swisher and NYU Professor Scott Galloway offer sharp, unfiltered insights into the biggest stories in tech, business, and politics. They make bold predictions, pick winners and losers, and bicker and banter like no one else. From New York Magazine and the Vox Media Podcast Network.