Underground structures are one. Spacecraft are another. In each case, the principle problem is that the usual methods of rapid heat transfer --- conduction, convection, and radiation --- are constrained. And when heat accumulates, it's a considerable challenge.
The one fairly obvious thought that occurs to me is that this poses a potential opportunity for winter-time heating via ground-loop heat pumps. So long as there's a substantial amount of heat being transferred into the ground, it might also be extracted.
Cities are often known for district heating and cooling, where entire neighbourhoods have centrally-provisioned heat (often residual steam from electrical generation or industrial processes). I'm not sure that low-grade ground thermal heat could readily be transported (say, from Chicago's downtown "Loop" to surrounding neighbourhoods) in the same way, but this does seem an potential option.
There's also nearby Lake Michigan which might serve as another potential cooling reservoir should active chilling of the ground prove necessary, though that would be a major engineering undertaking of itself.
I've some suspicions about the study methodology which relies on sensors placed within structures, and in one case literally in a boiler room, which ... would seem likely to experience anomolous and localised heating. (The lack of thermal monitoring within the Chicago subway system seems a lost opportunity.) I suspect that a system of ground wells sampled throughout the city might provide a more robust set of data on actual ground-temperature patterns over time.