Uber’s Workforce Embraces Dara AI Technology

Today, I’m chatting with Uber’s CEO, Dara Khosrowshahi. It’s become a yearly ritual for Dara to swing by the studio during his visit to New York for Uber’s GO-GET event, and it’s always a blast.

This year, the big news is Dara’s vision of Uber evolving into a comprehensive travel platform. The app will now allow users to book hotels, thanks to a collaboration with Expedia. New features are also rolling out, like having coffee and snacks available in your Uber upon arrival, and even personal shopping services. Uber is dubbing this an “everything app,” so I wanted to explore how far Dara envisions this expansion going and whether he feels the pressure to enhance the user experience, especially as AI companies claim their chatbots will handle all your transportation needs.

I was curious if these chatbots have opened any doors for Uber. Last year, Dara mentioned he was open to partnerships to explore meaningful integrations, but the AI features I’ve seen so far have been clunky and slower than just using the app. We discussed what Dara is observing in this space and the potential for future developments.

I’ve also been eager to hear from software CEOs about how AI is reshaping their companies, especially as AI coding tools and autonomous systems revolutionize software development. Recently, Uber’s CTO revealed that the company had exhausted its entire AI token budget for the year by early April, prompting Dara to reconsider hiring strategies as expenses on tokens rise. This is a significant gamble, and I wanted to know if Dara is reevaluating how his software teams are structured as AI continues to blur the lines between product managers, designers, and engineers.

We also delved into Uber’s substantial investments in autonomous vehicles, particularly its major stake in Rivian, discussing the milestones Dara is looking for as the technology progresses and what the future holds for Uber drivers in a world increasingly dominated by robots.

Finally, I asked Dara when he thinks AI will be ready to take over the CEO role. Interestingly, there’s already a rogue AI version of Dara within Uber.

There’s a lot to unpack here—Dara was as straightforward and open as ever, and I think you’ll find it engaging.

Alright, let’s dive in: Uber CEO Dara Khosrowshahi. Here we go.

This interview has been lightly edited for length and clarity.

Dara Khosrowshahi, welcome back to Decoder.

Thanks for having me back!

It’s great to have you; it’s like a yearly tradition. You come down after your GO-GET event, and there’s always a lot of news to discuss.

Absolutely, we’ve got plenty to share!

I want to kick things off by discussing the new feature that allows users to book hotels and experiences through the Uber app. But first, I always ask the same two Decoder questions about company structure and decision-making. Let’s do a quick lightning round.

Last year, you mentioned your decision-making framework, referring to “one-way doors and two-way doors.” With the recent expansion of the app, has your approach to decision-making evolved?

The core framework remains unchanged. However, I’m pushing the company to embrace smart risks. I’ve noticed that as companies grow, they often become more risk-averse, focusing on playing it safe. While being a public company means we have to meet quarterly expectations, we also have the resilience to take bigger risks. When I joined, a billion-dollar mistake could have been catastrophic, but now, with nearly $10 billion in cash flow, we can afford to take calculated risks to drive innovation, especially in areas like autonomous vehicles.

We need to be willing to take those risks, even if some decisions may seem unwise in hindsight. It’s crucial to push boundaries, especially in this era of rapid innovation.

Taking risks is one thing, but accepting responsibility for failures is another. How do you empower your team to take risks without the fear of failure?

If you can’t identify the downside, don’t take the risk. But if you can, whether it’s time spent on a feature or capital invested in a new business line, you can make informed decisions. We want to learn from our mistakes, but I’m not one to celebrate them. Instead, I focus on ensuring we learn from them to improve future decisions. Sometimes, we overanalyze our mistakes, but I prefer to understand what went wrong and move forward to build the next thing.

Can you share an example of a risk that was outside your control but turned out well, and one that didn’t?

One successful initiative was the “women riders and drivers preferred” feature. Initially, there were concerns about marketplace liquidity, as most of our drivers are men. However, we managed to build a reliable pool of women drivers, creating a positive feedback loop.

On the flip side, we attempted to build a taxi product twice. The first attempt failed because taxis didn’t trust us. Years later, we tried again with a different approach, implementing blast dispatch to improve reliability. This time, it worked, and taxis are now one of our fastest-growing products. It’s a reminder that sometimes you need to revisit ideas with a fresh perspective.

We’ll circle back to risk later, as you have several new products that seem ambitious.

Let’s talk about Uber’s structure. Last time, you described it as a combination of matrix and line of business structures, with product as a central function. Has anything changed with the addition of new lines of business?

The only structural change is the appointment of Andrew Macdonald as president and COO. He previously ran our global mobility business. We’ve observed that users who engage with both mobility and delivery are growing faster than those who use them separately. My vision was that once delivery grew post-COVID, it could rival mobility. Competing against both mobility and delivery players is challenging, so we needed a unified approach.

We’ve been pushing the idea of a consumer-side and driver-side platform. The more we can get users to engage with both Rides and Eats, the better our retention and spending become. We launched the Uber One membership, which is growing rapidly, and members tend to spend three times more than single-platform users. This strategy is working, and I want to lean into it.

It sounds like you’re navigating a lot of trade-offs in this process.

Everything is a trade-off. Andrew manages those trade-offs, ensuring we balance customer satisfaction with profitability.

You mentioned that understanding your organizational structure can reveal a lot about a company’s challenges. It seems you’ve appointed someone to oversee prioritization and trade-offs more holistically.

Yes, Andrew’s role is crucial for managing those trade-offs effectively. He’s responsible for customer satisfaction and profitability, which often intersect.

I’ve noticed that AI companies are in a different position now than they were a year ago. They’re racing toward IPOs and facing intense scrutiny. How do you view your relationship with AI companies as you build out Uber’s AI capabilities?

We’ve had a strong partnership with OpenAI, and their products have been excellent. We’re using their models for various applications, and despite market drama, their quality remains top-notch.

As we look to the future, we’re focused on building AI services that enhance user experiences. We want to create moments of surprise and delight, making it easier for users to interact with Uber.

Let’s shift gears to your investments in autonomous vehicles. You’ve committed over a billion dollars to Rivian for robotaxis. Can you clarify the nature of this investment?

We’ve invested in Rivian’s equity and committed to purchasing vehicles. This partnership is both an investment and a vehicle commitment, contingent on achieving specific milestones.

What kind of milestones are you looking for?

While I can’t disclose specific milestones, they generally involve in-market releases and safety benchmarks. The industry has moved beyond rigid level systems; we’re focused on practical capabilities and economics.

You’ve diversified your partnerships beyond Waymo, which has been a significant player in the autonomous space. What’s your strategy there?

We believe multiple players will emerge in the autonomous vehicle market, and we want to be positioned to leverage various partnerships. Our goal is to build a robust ecosystem that supports these companies in achieving superhuman safety.

You seem confident that autonomy is on the horizon, which contrasts with the typical “10 years away” narrative we’ve heard in the past.

All evidence suggests that progress is accelerating. Waymo has set a precedent, and we’re seeing advancements from various players. We’re committed to investing in this space to ensure we’re ready when the time comes.

As we discuss the implications of autonomy, what do you foresee for Uber drivers in the future?

In the next decade, I’m confident we’ll have more drivers on our platform than today. We’re also exploring new roles for drivers, such as personal shopping, to adapt to the changing landscape. While I can’t predict the long-term future, I believe we’ll find ways to integrate human drivers alongside autonomous vehicles.

Finally, I want to address a question that many Uber drivers have: How can they earn more?

Driver pay is influenced by market rates, and we’re working to create new opportunities for drivers. As we introduce more complex roles, we hope to increase their earning potential.

Thank you for your insights, Dara. It’s always a pleasure to have you on the show.