Investigating the interaction of people with their environment, how humans shaped and transformed it in the long-durée, has long been a trajectory in archaeology. Moreover, thanks to computational methods and machine learning techniques nowadays widely adopted and developed in archaeology, the possibilities to explore, interrogate and disentangle archaeological information in order to simulate and better understand the past, have significantly increased.

The aim of this talk is to give an overview of the multiple and infinite uses of computational approaches and machine learning techniques in archaeology. Synthesising data and results from projects carried out in the past years, and drawing upon different materials, several case studies ranging from the Mediterranean, Asia and South America, will be presented.

On the one hand, the integrated use of multi proxy and machine learning methods in a paleo-ecological and paleo-climatic framework analysing the dynamics of the first, small-scale Neolithic farming societies facing climate constraints in the Mediterranean, and on the other the complementary methodological application of a multi-agents based model to simulate Neolithic hunting strategies within and around the desert kites structures in Saudi Arabia will be tackled. Finally, the talk will summit on the on-going research project – ADArchaeoSA – in South America, to explore the most cutting edge solutions to analyse complex patterns of pre-Inca landscape occupation and transformation in the Andean highlands, by relying on deep-learning based site detection workflows.

It will be argued here how these methods have become increasingly valuable and needed in the discipline for describing and navigating the challenges of past and present questions of human-environment complexity, developing multi-proxy modelling approaches, working towards data reproducibility and accessibility, and dealing with bias and uncertainty.