AI Arms Race

The Risks of an AI Arms Race

• AI can now perform tasks such as writing, coding, art and predicting protein folding.
• Many AI researchers believe that high-level machine intelligence could have a 10% chance of leading to "extremely bad" impacts for humanity.
• On May 30, hundreds of AI scientists and CEOs of top AI labs signed a statement urging caution on AI.
• The simplest argument against AI is that it could lead to the creation of superhumanly-smart artificial entities with goals that conflict with humanity's interests.
• If a safer lab pauses, a more reckless lab could take over and put our future at risk.
• The AI situation can be analogized to a classic arms race, but with AI, the winner may be advanced AI itself.
• Other game-changing factors in AI include how much safety is bought by going slower, whether coming second means a small loss or a major disaster, and how others respond.
• In the real world, we can coordinate our way out of such traps by talking to each other, making commitments and observing adherence, and lobbying governments to regulate and make agreements.
• The payoffs for a given player in AI can be different from the payoffs for society as a whole.
• The narrative of AI as an arms race strengthens the interests of researchers and investors chasing fame and fortune.
• The best individual action in AI could be to move slowly and cautiously, and collectively, we shouldn't let people throw the world away in a perverse race to destruction.
• Coordinating our escape from the risks of an AI arms race has scarcely been explored.