I would strongly (>95% confidence) predict no. But not so much because there are no general rules but because the riddles are basically reverse-engineering algorithms aka riddles.
I wonder whether the ability to solve these kinds of riddle games is a fire alarm for AGI or at least another bigger milestone.
Would any current reinforcement learning algorithm be able to solve this game?
I would strongly (>95% confidence) predict no. But not so much because there are no general rules but because the riddles are basically reverse-engineering algorithms aka riddles.
I wonder whether the ability to solve these kinds of riddle games is a fire alarm for AGI or at least another bigger milestone.