AI has very quickly made its way into our lives with an endless number of free and powerful tools on the market. Its applications have had a marked impact on medicine, education, software, and even video game development. The pace at which AI is progressing and improving can be a bit disconcerting, but we might just be underestimating the danger it poses. Elon Musk has shared his views on AI since the beginning, and according to him, there is a 10-20% chance that AI will take over humanity.
This may seem like science fiction or fear-mongering, but is not the case. AI is one of the biggest leaps in technology in recent memory, and according to Elon, it won’t be powerful enough to be a danger to society until 2025. But during the Abundance Summit, he stated that “there is some chance that it will end humanity.”
Elon Musk Gives AI a 10-20% Chance of Ending Humanity, While Others Believe It’s Far Worse
But oddly enough, even with such gloomy predictions, Elon believes we should continue exploring and advancing AI technology. He thinks that the likelihood of something good coming out of it is more important than the off chance it destroys humanity. At worst, we’ll just cause the entire extinction of the human race, no big deal.
On the other hand, the AI safety researcher of the Cyber Security Laboratory at the University of Louisville, Roman Yampolskiy, paints a much darker picture. According to Yampolskiy, Elon’s estimate of 10-20% is far too conservative, and the reality will be much worse. He believes the risk is exponentially higher and also referred to as “p(doom)”, which refers to the likelihood of AI destroying humanity.
Everyone who understands and believes in (p)doom believes the risk to be extremely high, while Yampolskiy said to Business Insider that the chances of AI ending humanity as we know it is 99.999999%.
Elon Musk, among many others, proposed a slow-down in AI development when GPT-4 was nearing completion. Other than that, the only option is to work on AI using segregated tools that can’t result in the creation of an AGI (artificial general intelligence).