Uber's AI Push: Transforming Code with Minimal Human Touch

Uber, once a disruptor in how we hail rides, is now making waves in how it produces software. The company's Chief Technology Officer, Praveen Neppalli Naga, announced that Uber's AI agent is now responsible for 1,800 code changes each week. It's a bold move, reflecting a broader trend across industries where AI is reshaping the software engineering landscape.

AI in the Driver's Seat

Naga highlighted that 95% of Uber's engineers are engaging with AI monthly, marking what he calls a 'real reset moment for engineering.' The move to lean heavily on AI is evident as Uber's internal coding agents autonomously generate code with zero human authoring. Engineers review and approve, but the heavy lifting is done by AI. This isn't just a tweak. it's a seismic shift in software production.

Agentic AI, where software independently executes tasks, is at the core of this transformation. Uber has seen its AI increase its share of total code changes from less than 1% to 8%. However, the deployment actually looks like a blend of innovation and necessity, as more companies explore AI to boost productivity.

The Broader Implications

Uber's approach is part of a wider movement. A study by engineering intelligence platform Jellyfish reveals that 63% of companies now use AI for most of their coding. Enterprises don't buy AI. they buy outcomes, and for Uber, the outcome is faster code development.

Yet, it's not all smooth sailing. Amazon's experience serves as a cautionary tale. Following a series of outages tied to its AI coding tools, Amazon had to reinforce its systems to prevent operational disruptions. The real cost of AI isn't just in implementation but in the potential pitfalls of rapid, unsupervised changes.

A New Role for Engineers?

For software engineers, the ground is shifting. The work now involves more system architecture and code review than line-by-line coding. While this might seem like an efficiency gain, it raises questions. Is this the future engineers signed up for? Some report increased workloads leading to burnout, as they adapt to new roles shaped by AI's presence.

Despite the challenges, the adoption curve continues to climb. Uber's strategy of encouraging engineers to 'quietly experiment' with AI, rather than enforcing adoption from the top, might be the key to smoother integration. As AI coding becomes more prevalent, companies will need to balance innovation with the risks of over-reliance. The consulting deck says transformation. The P&L says different.