Twenkid, do you agree that technology has improved since 2010? Do you think it will continue to improve?
You can see what predictions I got right and what I got wrong in my 1999 paper on text compression as a test for AI and my 2008 distributed AGI proposal. http://mattmahoney.net/AAAI99-177.pdf http://mattmahoney.net/agi2.html In 1999 I argued that all you need to pass the Turing test is text prediction. That happened in 2023. In 2008 I predicted that AGI sufficient to automate human labor would cost $1 quadrillion and require public acceptance of surveillance on a massive scale. We have the surveillance. We have tens of trillions already invested in AI with barely a dent in the job market yet. The world economy is booming as I predicted it would. I got Moore's law wrong. I assumed it would make hardware costs negligible. I did not anticipate clock speeds would stall around 2-3 GHz in 2010 or that transistor sizes would stall around 20 nm in 2025. But I think we are proof that we can go beyond this with nanotechnology. DNA bases are a million times heavier and slower than electrons, which makes them a million times more energy efficient because E = mv^2/2. But I don't know how long it will take for nanotechnology to catch up to transistors. My design focus was on efficiency. It was based on my 1998 thesis showing that a global distributed search index would have roughly O(log n) access time and O(n log n) storage, even with rogue peers in a hostile environment. It would be funded as peers compete for attention in an environment where information has negative value. I did not give any thought to using a peer to peer network to evade censorship because it was hardly an issue in 2008. I thought that China would benefit by dropping its Great Firewall, not imagining that every other country would instead adopt the technology they pioneered. It wasn't until the Arab Spring of 2010 that governments recognized social media as a threat. The fatal flaw in my distributed design is that it did not overcome the advantages of centralization, with the subsequent concentration of power. We shop on Amazon instead of a thousand different websites because we trust it. We use Google maps instead of a thousand different apps because live traffic updates wouldn't work otherwise. We put security cameras in our house that capture us walking around naked and don't mind because random people aren't going to see it like they would in my design. In my design, all data is public including queries, so nobody can stalk you without you knowing and stalking them back. But lacking this, we are still OK with machines hearing every word, watching every move, and tracking every dollar, because they make our lives more convenient and they can't do anything illegal or creepy with that data. -- Matt Mahoney, [email protected] ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T574b10b838cba731-M512131886a60c88757441917 Delivery options: https://agi.topicbox.com/groups/agi/subscription
