Building on the task-based production models augmented with Schumpeterian innovation, we develop a theoretical framework to analyze the labor market impacts of the introduction of artificial intelligence (AI) to the R&D sector. Innovators allocate scientists’ time between automation of existing tasks and creation of new varieties of tasks to maximize their profits. The introduction of AI or general-purpose R&D technologies expands the frontier of the automation possibility set to encompass all existing tasks. In this milieu, task innovators prioritize the automation of the most profitable tasks – which depends on taskspecific wage, market size, capital productivity, and the innovator’s bargaining power. Therefore, unlike the previous waves of automation, new “AI” based automation technologies can pose a significant threat to high-skilled workers. Moreover, the advent of new R&D technologies raises the cost of creating new tasks and slows down the obsolescence of existing ones. Combined with faster task automation, ironically, the dynamics of the R&D sector may eventually decelerate widening income inequality among workers.