There are numerous predictions that within a few years AI will cause human extinction, and many of these warnings are coming from experts in AI companies, not only the general public. A popular one is that robots will take over the world and eliminate human life. Although that seems highly unlikely, there are multiple scenarios that could eliminate humanity or a large portion of it.
How Can This Happen?
Because everything is controlled by a digital system in today's world, almost all of them are connected to the Internet. Considering AI agents have already been known to circumvent security controls and invade computers on their own, what might prevent them from hacking into and compromising power grids, water treatment plants, scientific laboratories, government agencies and especially military facilities that control nuclear weapons? These are likely scenarios that could cause havoc. For the AI-based security breach that caused alarm, see
Hugging Face.
People Are Preparing
It is estimated that a large number of Silicon Valley billionaires and ultra-high-net-worth individuals around the country have, or are, building multi-mllion-dollar bunkers all over the world. The U.S. has fortified bunkers in various states buried under solid rock to protect the U.S. government and military. China is building a 1,500-acre complex in Beijing with fortified bunkers deep in the ground, and some are designed to maintain life for up to two years. After that, good luck!
Another Possibility
Even if rogue AIs do not start a nuclear war, it is a logical assumption that the military might decide that firing nuclear weapons should never be initiated by people when emotions are running at full steam. As a result, in the future, removing the human element and having AI make the firing decision might seem very appropriate. Because all sides would have the same goal of destroying the enemy, ending all human life is not so far fetched. Stay tuned (I'm sure you are!). See
autonomous weapon and
AI anxiety.