I will try to reproduce these two papers that suggest program synthesis (eg: inducing new tactics) for AI to solve math problems or the ARC-AGI puzzles: • https://royalsocietypublishing.org/doi/epdf/10.1098/rsta.2022.0044 • https://arxiv.org/abs/2407.00695 There's a quite a bit of new math/ML/programming for me to learn here.