How can AI cause humanity to perish?

(Dan Tri) - Experts warn that AI can go beyond human control, pursuing its own goals, causing serious consequences, and in extreme scenarios, it can lead to the extinction of humanity.

In early September, Jacob Coxon, an AI expert who worked at OpenAI and Anthropic, two leading artificial intelligence development companies today, attracted public attention when he posted a message warning that AI could kill humanity by the end of this decade.

"The people who are building AI truly believe it could kill us all by the end of this decade. They are rushing towards self-improving artificial super intelligence and gambling with our lives," Jacob Coxon wrote on his personal page.

Less than 72 hours after the article was posted, the post had more than 164 million views. Many AI employees and industry leaders have supported Jacob Coxon's views.

Artificial intelligence expert Jacob Coxon caused a stir when he warned about disasters created by AI (Photo: CBS).

Evan Hubinger, another Anthropic employee, also voiced support for Jacob Coxon's views and estimated that the possibility of human extinction caused by AI within the next 10 years is greater than 10%.

"Basically, everyone on the planet will die. People often don't believe it when they say this, but I actually think that's the most likely scenario," Nate Soares, President of the AI ​​Research Institute (MIRI), commented on the message shared by Jacob Coxon.

Nate Soares is also the co-author of the book "If Someone Builds It, Everyone Will Die: Why Superior AI Will Destroy Us" which outlines the threats to humanity once superior artificial intelligence appears.

In fact, this is not the first time there have been warnings about the risk that artificial intelligence could cause humanity to perish.

In 2014, when artificial intelligence was still in its infancy and there were no powerful AI tools like today, billionaire Elon Musk warned about the possibility that AI could be more dangerous than nuclear weapons in the future.

"We need to be very careful with artificial intelligence. It is possible that it will become more dangerous than nuclear weapons," Musk posted a warning message on his personal Twitter page (now Twitter).

Elon Musk once expressed concern and compared AI to being as dangerous as nuclear weapons (Illustration: GPT).

Despite expressing concern with artificial intelligence, in December 2015, Elon Musk was one of the founders of OpenAI, the world's leading AI company today.

After founding OpenAI, in 2017, Elon Musk once again warned that artificial intelligence could be the cause of World War III, as major countries raced to develop AI systems to compete with each other.

Emmett Shear, who served as interim CEO at OpenAI, said he wouldn't be surprised if AI in the future becomes smarter, threatens humans and overcomes shutdown orders.

"AI today is relatively weak, but they will become much smarter and more powerful. It will not stop," commented Emmett Shear.

Professor Geoffrey Hinton, who is considered the "godfather of AI" for his contributions to building the foundation for the development of artificial intelligence (AI), has warned that there is a 10% to 20% chance that AI will wipe out humanity.

"Artificial intelligence will be much smarter than us. They will have countless ways to overcome control," Geoffrey Hinton shared in an interview in 2025.

Professor Hinton warns that in the future, AI systems can control humans as easily as adults can use candy to seduce a 3-year-old child. Hinton gave the example that AI systems are now willing to lie, cheat, and give false information to achieve their goals.

“AI Godfather” Geoffrey Hinton regrets not setting safety rules early when laying the foundation for AI development (Photo: Futura).

According to Hinton, instead of forcing AI to obey humans, he believes that the only solution for humanity to survive against AI is to integrate "maternal instinct" into AI models, so that they truly care about humans, even when AI has become stronger and smarter than the intelligence of all humanity.

“If they become more intelligent than humans, AI systems will quickly develop two goals: One is to survive and the other is to find ways to gain more control,” commented Geoffrey Hinton.

He argues that this is why it is important to integrate and nurture “maternal instincts” within AI models. However, Hinton said he is still not sure how to do this technically, but emphasized that computer scientists need to find a solution to make it happen.

Professor Hinton also expressed regret because he only focused on helping AI work without paying attention to other issues when starting to lay the foundation for this science, showing that he was really concerned about the possibility of AI going beyond human control.

Sharing the same opinion as Professor Geoffrey Hinton, Paul Christiano - a pioneering computer scientist in AI safety and alignment research - also expressed concern about the development of artificial intelligence and said that if AI surpasses human control, it will be a disaster for all humanity.

Leading expert on artificial intelligence Paul Christiano fears disaster for humanity when AI goes out of control (Photo: Dwarkesh).

"I think there will be about a 50% chance of AI surpassing human control and becoming a disaster for humanity. There will be about a 10 to 20% chance of AI becoming a deadly disaster for humans," said Paul Christiano.

According to Paul Christiano, artificial intelligence, like a newborn child, can develop on its own based on its own "life experiences". Just like a baby will cry when hungry so its parents respond by breastfeeding, the AI ​​system will also try to achieve its given goal by trying many different solutions until it achieves the result.

Therefore, Paul Christiano believes that once AI sets a goal of overcoming human control, it will find a way to do it.

To be continued…