Will AI kill us all

AI researchers estimate a 10% median probability that future AI advances could cause human extinction or permanent human disempowerment, up from 5% a year earlier, according to a December 2024 survey of 1,580 top AI conference publishers. However, expert forecasters with strong track records give lower estimates (2.4-5%), while the general public shows greater concern, with 50% of US adults worried AI could end humanity. The timing of potential AI risks has shifted significantly, with researchers now predicting human-level AI could arrive by 2042 instead of the previously estimated 2061.