Jeff Hawkins on neuromorphic AGI within 20 years — LessWrong