MIT study finds 21.5% AI catastrophe risk by 2030
Technology
A new MIT study says there's a 21.5% chance that artificial intelligence (AI) could trigger a worldwide catastrophe by 2030, specifically tied to the highest-ranked concern of AI systems developing dangerous capabilities: think more than 1 million deaths or $100 billion in damage.
Researchers warn AI gaining dangerous capabilities
The biggest fear is AI gaining dangerous skills (like hacking, spreading fake information, or even developing weapons) with a 21.5% chance of causing huge trouble.
Other risks include powerful cyberattacks and unfair power grabs that could make society less equal.
Even with safety measures in place, the researchers say we need serious teamwork across countries to keep things safe, with Peter Slattery noting that there are many AI risks.