An Artificial Intelligence Arms Race
https://www.theverge.com/2017/9/4/16251226/russia-ai-putin-rule-the-world
Earlier this month, Russia’s president Vladimir Putin said that whoever becomes the leader in Artificial Intelligence technology (AI) “will become the ruler of the world.” James Vincent of The Verge covered his statement and described a world in which AI is used in warfare. Putin predicted that future wars will be fought using drones and The Verge further speculated that these drones will be powered by AI.
Elon Musk, the chief innovator behind world changing companies like Tesla and SpaceX, has been a vocal advocate for AI regulation. He responded to The Verge’s article in a tweet that said, “…Competition for AI superiority at a national level [is the] most likely cause of WW3 imo [in my opinion]”. He added that a sophisticated AI, built to determine the most probable path to victory, could decide to place a strike without permission from a country’s leader.
Will there be an arms race for AI? Let’s consider a very simple table with arbitrary numbers representing the potential payoffs and strategies of two countries:
Country A ↓ Country B → | Don’t develop AI weaponry | Develop AI weaponry |
Don’t develop AI weaponry | 4, 4 | 1, 5 |
Develop AI weaponry | 5, 1 | 2, 2 |
Here, the payoff when both countries do not develop is greater than the payoff when both countries develop. But the best outcome for country A is a payoff of 5 when A develops and B does not. B receives the worst payoff possible in this case. Hence developing AI weaponry is the dominant strategy.
This simplified version is missing the multitude of AI applications beyond weaponry. Companies are using IBM’s AI technology to diagnose lung cancer, explore space, and combat joblessness.
Here’s a new version of our chart considering all AI applications:
Country C ↓ Country D → | Don’t develop AI | Develop AI |
Don’t develop AI | 4, 4 | 1, 5 |
Develop AI | 5, 1 | ?, ? |
If country C does not develop AI while country D develops, country C will struggle to compete in the global market and receive a payoff of 1. What numbers belong in the bottom right? Do the merits of AI outweigh the possible consequences? Will AI present a case of Prisoner’s Dilemma or not? Either way, it is in the United States best interest to invest in computer science education and AI research.