Superintelligence: the ultimate geopolitical weapon?

Description

The US and China are in a race for advanced AI, specifically superintelligence, an AI that is much smarter than a human. This race is occurring in the private sector between many different AI labs, but because superintelligence can grant powerful military and economic dominance, governments themselves are likely to get involved. However, superintelligence is likely to concentrate power amongst a very small group of people. Through taxation and separation from resources, it's likely to impoverish other countries, who have a strong incentive to prevent the creation of superintelligence. We may see a range of responses from espionage to physical attacks on data centers to delay superintelligence projects. Worst of all, nobody has ever built superintelligence before. So we don't know how difficult it is to control. Every superintelligence project has a risk of losing control, and the outcomes from this could be catastrophic for humanity. In the worst case, everybody could die as a result. This will only strengthen the desire of other countries to prevent the creation of superintelligence, increasing the risk of conflict between nations. #superintelligence #geopolitics #existentialrisk Take Action | ControlAI https://controlai.com/take-action How to Hide an Empire: A History of the Greater United States https://www.amazon.ca/How-Hide-Empire-History-Greater/dp/0374172145 A Memory Called Empire https://www.goodreads.com/book/show/37794149-a-memory-called-empire Superintelligence Strategy https://www.nationalsecurity.ai/ Superintelligence Strategy [paper] https://files.nationalsecurity.ai/Superintelligence_Strategy.pdf If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All https://www.amazon.com/Anyone-Builds-Everyone-Dies-Superhuman-ebook/dp/B0DZ1ZTPSM AGI Ruin: A List of Lethalities https://www.lesswrong.com/posts/uMQ3cqWDPHhjtiesc/agi-ruin-a-list-of-lethalities Colonialism (Stanford Encyclopedia of Philosophy) https://plato.stanford.edu/entries/colonialism/ An intellectual history of digital colonialism https://academic.oup.com/joc/advance-article/doi/10.1093/joc/jqaf003/8078024 AI dominates venture capital funding in 2024 https://www.fdiintelligence.com/content/41641e67-f00f-53c0-97cb-464b3a883062 Missile Defense | The Heritage Foundation https://www.heritage.org/military-strength/assessment-us-military-power/missile-defense 0:00 Intro 0:23 Contents 0:31 Part 1: The US-China race for superintelligence 0:46 Superintelligence definition 0:59 We are building superintelligence 1:17 Powerful AI will lead to economic dominance 1:50 Why governments aren't directly involved (yet) 2:04 What the US government is doing 2:24 Superintelligence can lead to military dominance 3:01 What the Chinese government is doing 3:24 US export restrictions for Nvidia GPUs 4:21 Part 2: Dominance or catastrophe 4:41 Other countries want to stop superintelligence 5:17 Example: Winning the Civilization game 5:51 Aside: Colonialism and capitalism 6:30 Aside: US is seat of global economic empire 7:05 AI will lead to digital colonialism 7:53 How can we control superintelligence? 8:27 Risk of existential catastrophe 8:51 Could there be a goldilocks accident? 9:21 Materials for further reading (new book) 9:45 Part 3: Geopolitical response 9:58 Existential risk to other countries 10:29 Superintelligence strategy paper 10:49 MAIMing a superintelligence project 11:10 Existing AI research hotbeds and vulnerabilities 11:49 Even the US is taking a risk by building superintelligence 12:14 Four possible paths forward 12:25 Path 1: Default path is not good 12:41 Path 2: Control the development of AI 12:57 Path 3: Centralize AI development 13:39 Path 4: Stop AI development entirely 14:09 An AI accident might make this more palatable 14:30 What will actually happen? 14:54 Developing superintelligence is a fool's errand 15:20 ControlAI tool to contact representative 15:44 More conversation needed 16:05 Conclusion 16:43 What we don't want 17:25 Outro