Develop AI systems to efficiently learn new skills and solve open-ended problems, rather than depend exclusively on AI systems trained with extensive datasets
In reasoning, it seems to me that there are multiple resolutions or granularities we can take as we work to solve the problem.
This would be the first step, where we learn what kind of task we need to reason about.
Is is a translation, recoloring, restructuring, shading, shifting, etc. task? This is what we'd narrow down with the first level of problem solving, then continue the task at finer granularity to find a more exact answer.
If we could use our encoding of the input with varying resolution, potentially we could use this to continuously approach the problem from different levels of specificity.
Discover other posts like this one