Can we accurately estimate pdoom?
Sources:
Calculating the probability of a catastrophic event, commonly referred to as "P(doom)," is considered highly challenging by experts. has expressed skepticism about such calculations, deeming them as "very sloppy" due to the need for a stochastic path integral through all possible futures, acknowledging our biological tendency toward biased negative outcomes based on evolutionary history 1. He emphasizes that the space of all possibilities is extremely vast, and chaos theory further complicates the prediction of the future.
Other experts provide different perspectives, not necessarily solidifying a precise estimate but acknowledging the inherent challenges and risks associated with advanced technologies like AI. somewhat agrees with the ‘AI takeover scenario’, considering there to be something like a 10-20 percent chance of AI takeover, but suggests that there are also other potential dangers associated with AI advancement which could bring the overall chance of doom closer to 50 percent 2. On the more optimistic side, (referenced secondhand in a conversation) estimates a much lower probability of doom, around 10 percent 3.
In summary, while some experts express specific probabilities, others argue for the inherent difficulties and uncertainties in making such predictions. The discussions reflect a broad range of views emphasizing the complexity associated with estimating existential risks.
RELATED QUESTIONS
