Carbon fiber reinforced polymer (CFRP) has been used in many fields especially aerospace due to its many advantages over traditional materials. However, drilling CFRP creates many critical problems such as drilling defects and delamination. Developing analytical modelling reduces experimental tests number and the cost of defining optimal experimental parameters. In this work, drilling quality of woven CFRP, depending on cutting parameters and tools geometry, has been investigated experimentally. The results have been utilized to model the drilling process using response surface methodology (RMS) and artificial neural network (ANN). The analysis of variance (ANOVA) shows that, regardless the utilized drilling tool, entrance delamination and thrust force and torque evolution depend especially on feed rate value. However, exit delamination depends on feed rate for step and spur drills and spindle speed and feed rate for twist drill. RMS and ANN results have been compared. ANN is closer to the experiment than RMS. A multi-objective genetic algorithm (GA) was implemented to optimize the delamination’s apparition in drilled holes. The obtained results were arranged and analyzed to make the right decision in different drilling preferences. The use of spur drill provides optimal drilling quality and productivity. The interest of this study is to create an artificial intelligence program allowing the knowledge of optimal drilling conditions without doing experiment.