While the concern about tech CEOs having nefarious motives is very real, l just feel like I’m hearing increasing concern from academics with no profit or power motive. The concern mostly surrounds the reckless drive to create more advanced models with little to no attention paid to how to mitigate potential dangers.

These models are going to improve significantly over the next couple years given the current rate of advancement, and the people training these models genuinely do not understand them any more than you can understand the inner workings of someone else’s mind. One of these getting loose onto the internet and potentially causing catastrophic interruptions or damage to any internet-connected system is not hard to imagine.