In the context of smart energy grids, demand-side management refers to the ability of dynamically controlling and scheduling energy-consuming tasks. In one potential deployment scenario, smart appliances are controlled by a local intelligent software agents, which implement a given optimization algorithm for scheduling such tasks. The higher the fraction of users adopting such technology, the higher the advantage for the energy operator, due to the ability to control load curve and smooth peaks. At the same time, single users may incur some penalties, related to the fact that energy-consuming tasks may be deferred, thereby causing inconveniences. In this paper we take a game-theoretical perspective at demand-side management techniques. Tools and solution concepts from evolutionary games are employed: we are interested in the dynamics of the adoption of demand-side management schemes by intelligent software agents. We focus on distributed control schemes that can be enforced by the operator through pricing schemes. Agent-based numerical simulations are provided to validate our theoretical results.