In this article, Ramez Naam argues against a fast takeoff. He notes that the recursive self improvement loop is only 5 to 10% as strong as it needs to be for automatic gains and that performance and productivity only increase with the log of computing power. Computing power is increasing exponentially, of course, resulting in a linear rate of increase. He did not give a predicted time until fully automated takeoff. But from reading the paper it looks to me decades away. https://open.substack.com/pub/noahpinion/p/wheres-the-intelligence-explosion
Meanwhile I was watching Arthur C Clarke's 1964 BBC video where he essentially predicts the Internet, AI, and machines that evolve thousands of times faster than biological life. He doesn't see human extinction as a disaster, but rather a stepping stone to something better. He didn't give a date either, knowing how hard it is to predict the future. I estimated 130 years at the current rate of Moore's law to surpass the computing capacity of the biosphere, but who knows if the trend will continue that long? -- Matt Mahoney, [email protected] ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Td23f8d2a801e2642-M43cc54cb8b772d505e7421ce Delivery options: https://agi.topicbox.com/groups/agi/subscription
