The Extremely Large Telescope (ELT) is currently under construction. When completed it will be the world's largest optical/near-infrared telescope, providing unprecedented high angular resolution images and enabling new astronomical revelations when combined with a range of cutting-edge instruments.
To reach the diffraction limit and realise the full potential of the ELT each of the instruments will utilise Adaptive Optics (AO) systems. In this talk I will introduce some of the challenges associated with building AO systems for segmented telescopes at this scale. I will explore specific challenges in wavefront sensing from the perspective of the Single Conjugate AO system of HARMONI, one of the first-generation instruments. Finally I will discuss the areas in which new data driven machine learning techniques have the potential to improve AO performance, from dealing with issues of non-linearity for the most sensitive wavefront sensors, to taking advantage of new detectors which provide additional information about the incident photons.