Algorithms have long been the backbone of computing, but with the advent of artificial intelligence they are shifting from explicit occurrences to implicit instances, being just a bunch of weights in the model. But the question is: can AI do a better job than hardcoded algorithms?
I personally don't think AI models will ever outperform existing algorithms, but they are totally awesome at things for which we have no go-to algorithm yet. And they are good at finding more optimal ones: The routing Dijkstra routing algorithms was recently optimized furhter optimized by an AI framework. This advancement, detailed in a March 2026 arXiv paper, represents a shift from static route calculations to adaptive, context-aware planning, potentially cutting operational inefficiencies in global delivery networks.
