Can Large Language Models be leveraged to improve student learning outcomes? Specifically, can they provide sufficient individualized assistance to students learning how to program?