Will AI Really Wipe Us Out by 2030? - Nuclear Engineer Reacts to CNN

could make it do the tasks that we're currently doing, and then you get what's called an intelligence explosion. >> Ultimately, physics has to sign off on it. So, forget human constraints, you're going to run into physical constraints, no matter how smart your AI is. Today, we're going to be looking at an Anthropic employee's claim that AI can

unalive all of humanity by 2030. This is his CNN interview with Anderson Cooper. Let's see how much merit this claim has. >> Coxen has left Anthropic sounding the alarm about the dangers. Jacob Coxen has left Anthropic sounding the alarm about the dangers of artificial intelligence. On his way out the door, the 27-year-old

outlined his concerns in a series of social media posts when he quit, writing in part, "The people building AI earnestly believe that it could kill us all by the end of the decade." >> Starting off strong with a pretty extraordinary claim. Now, I'm not going to dismiss the claim because it sounds dramatic. After all, throughout my

career as a nuclear engineer, there's been a lot of assessments and analysis about low-probability, high-consequence events. The fact that an accident hasn't happened yet doesn't mean it can't happen, talking about this scenario, obviously. And the consequences of certain failures can be enormous. However, the severity of an accident by itself doesn't tell you how likely it is

going to occur. A nuclear power plant can experience a loss of coolant, a loss of all electrical power, equipment failures, operator errors, or some combination of those events. You don't just lump all of those possibilities together, announce that the consequences could be catastrophic, and call that a risk assessment. You build accident sequences. You identify the initiating