Amazon's AI boom is creating a mess of duplicate tools and data inside the company

Amazon is grappling with an emerging challenge as its AI initiatives lead to a surge in duplicated tools and data within the company. With various teams rapidly developing their own AI-powered applications to streamline workflows, the unintended consequence has been an increase in software and data overlap, as highlighted in an internal document obtained by Business Insider. The document reveals that the proliferation of AI tools is exacerbating the existing issue of tool duplication. "AI is making our tool duplication problem worse," it states, noting that the pace of duplication is accelerating while efforts to clean it up are lagging behind. This phenomenon is reflective of a larger trend in corporate America, where generative AI is giving rise to what some are calling "AI sprawl." This trend threatens to overwhelm companies' centralized oversight and security measures. As employees create tools in mere minutes using AI assistants, organizations may struggle to track which systems are active, where sensitive data is stored, and how much redundant software is in circulation. This situation echoes past challenges faced during the rise of cloud computing and the ensuing "SaaS sprawl," where employees utilized cloud resources without adequate oversight. Historically, companies eventually implemented strategies to manage such growth, and the same response is anticipated with generative AI, albeit at a much faster pace. Amazon, which has been at the forefront of AI integration under CEO Andy Jassy's guidance, is experiencing these effects more intensely. The company has acknowledged the problem of teams concurrently developing similar applications, a situation that was previously manageable but is now exacerbated by AI's ability to lower the barriers for creating new tools. AI enables rapid prototyping and deployment, allowing engineers to quickly build and maintain systems at a much lower cost. This has resulted in a proliferation of overlapping applications without the urgency to consolidate them. The internal document, produced by a team evaluating AI tools across Amazon's extensive retail operations, emphasizes that the situation is not just about tool overload. It also raises new concerns regarding data management. Many AI systems convert and store internal data in new formats, often creating additional copies that may not reflect changes to the original data. This disconnect can lead to outdated or incorrect information being accessed. In one instance, a system continued to display software details that had been made private in Amazon's internal code repository. As a response, Amazon is urging teams to document changes in permissions and data handling more meticulously. The rapid emergence of AI-generated tools is creating a landscape of "shadow AI," where unauthorized applications could expose sensitive information and lead to regulatory breaches. Experts warn that without proper governance, this could result in significant disruptions. To tackle the duplication issue, Amazon is exploring AI solutions that can identify redundant tools, flag potential risks, and encourage teams to consolidate their systems before overlaps become entrenched. However, the challenge lies in maintaining the balance between speed and coordination. Amazon's decentralized approach allows small teams to make swift decisions, which can fuel innovation but may complicate efforts to manage the growing complexity of their AI landscape.