Generative AI is rapidly entering knowledge work, promising major productivity gains – but what happens to the meaningfulness of work when AI takes over parts of human problem solving? This session looks at software development, where AI assistants like GitHub Copilot are already everyday tools. Drawing on a survey of 309 developers, we show that GenAI can improve both task performance and meaningfulness of work – but can also undermine them, depending on how it is used. The key mechanism is not simply effort reduction, but effort redistribution: GenAI can reduce unnecessary cognitive effort, while also supporting deeper learning and engagement when developers critically interact with AI. The takeaway for organizations: look beyond productivity metrics and design GenAI use that preserve learning, engagement, and meaningful work.