There are algorithms called ant colony optimization https://en.wikipedia.org/wiki/Ant_colony_optimization_algori.... They are modeled after this ant colony behavior. As others have mentioned, these are good at finding local optima, like tabu search or simulated annealing, or genetic algorithms. This is good enough for most business purposes, such as the 'couch production' case from the article and other business cases. However it is not the same as finding 'a general solution'. Sapolsky compares us being bad at finding 'a general solution' with ants capable of finding a local optimum. I find this a bit misleading.