We show how two-dimensional waveguide arrays can be used to probe the effect of on-site interaction on Anderson localization of two interacting bosons in one dimension. It is shown that classical light and linear elements are sufficient to experimentally probe the interplay between interaction and disorder in this setting. For experimental relevance, we evaluate the participation ratio and the intensity correlation function as measures of localization for two types of disorder (diagonal and off-diagonal), for two types of interaction (repulsive and attractive), and for a variety of initial input states. Employing a commonly used set of initial states, we show that the effect of interaction on Anderson localization is strongly dependent on the type of disorder and initial conditions, but is independent of whether the interaction is repulsive or attractive. We then analyze a certain type of entangled input state where the type of interaction is relevant and discuss how it can be naturally implemented in waveguide arrays. We conclude by laying out the details of the two-dimensional photonic lattice implementation including the required parameter regime.Introduction. Anderson localization (AL) [1], one of the most famous manifestations of quantum destructive interference, has been probed and verified in perhaps the most diverse physical platforms such as light propagation in spatially random optical media [2,3], noninteracting Bose-Einstein condensates in random optical potentials [4,5], microwave cavity fields with randomly distributed scatterers [6], and an integrated array of interferometers [7]. Interesting deviations in AL arise when interactions between the particles become significant. In fact, Anderson himself first noticed the importance of interaction in localization phenomena [8] and launched a theoretical investigation in collaboration with Fleishman [9]. Recently, advances in technology have reinvigorated theoretical [10, 11] and experimental [12-19] interest on this subject. For the special case of two interacting particles in a random one-dimensional (1D) potential, Shepelyansky has investigated the interplay of disorder and interaction and concluded that interaction modifies (weakens) localization [20]. However, several studies analyzing the two-particle case further [21][22][23][24][25][26][27][28][29] have shown that the problem of the interplay between disorder and interaction is very complex and the results depend on the details of the system, the localization measure, and the numerical technique employed [3,13,14].