One of the somewhat subliminal and entirely unsubstantiated assumptions we hear is that more computing power will allow us to approximate human intelligence — that quantitative augmentation will lead to qualitative improvements in automation. No new techniques are proposed — Just the scaling of existing ones.
When we study functional MRIs, we see that much of the brain is at rest at any given time. We read pop psychology articles about awakening the untapped intelligence, as if the brain volume at rest is a flaw. We notice what is lit up, but we tend to ignore the possibility of purpose behind inactivity. Here's the question these assumptions trigger.
Can not thinking be a form of intelligence?
Many can see the intuitive attractiveness of the Buddhist idea that clearing the mind can lead to enlightenment. We wish we could be more logical and more intuitive. It's clear that there is no consensus about how to balance these intellectual goals. If we admit this intellectual agnosticism, we are more likely to move toward the greater understanding of what intelligence can be at its best.
Is it possible that more computing power is necessary only for a small segment of problem types when approached under the assumption that power is necessary? Will these current approaches be later seen as primitive and limiting? Might a more enlightened approach we haven't yet considered lead to excellent performance on low end mobile phones for entire classes of problems?
What new ideas show promise along these lines?