Marvin Minsky gets at what is in my view the core problem with most mechanistic (and virtually all rule-based) theories of learning: " the kinds of AI projects that have been happening for the last 30 or 40 years have had almost no reflective thinking at all." Right. You can practice all you want, but if there is no means to allow this to actually change your understanding, the practice is of no use. But what does 'reflection' amount to in a rule-based system? Especially one where rules are 'innate' or hard-wired in the mind? If you are told what to do, or told what is true, where is the mechanism for changing this? Reflection (crucially) entails autonomy.
Via George Siemens.
Today: 4 Total: 20 [Direct link] [Share]
