The primary focus of this paper is the study of the effects that different Potential Energy Surfaces (PESs) have on the chemical kinetics of the O2+O system. A set of state-specific rate coefficients are computed by the Quasi-Classical Trajectory method (QCT) using the 9 PESs by Varga et al., and isothermal bath simulations are run. The computed QSS rates and relaxation times show good agreement with the existing literature. Further, the effect of different non-equilibrium processes is analyzed on the system relaxation. Finally, Neural Networks (NNs) and Gaussian Processes (GPs) are tested as alternatives for fitting the O2+O ab-initio points, and for generating PESs characterized by accurate energy predictions in the ranges of interest for hypersonic applications. The surrogate surface obtained through NN, in particular, is used for computing the state-specific rate coefficients, and for comparing mole fractions, global rates and relaxation times to the ones already in literature.