"Bomb artificial intelligence labs," two prominent scientists warn: They will bring the end of humanity on Earth

2025-09-26 19:09:13Kosova&Bota SHKRUAR NGA REDAKSIA VOX
"Bomb artificial intelligence labs," two prominent scientists warn:
Photo illustration

Two prominent scientists warn that the greatest threat to humanity may come from something much more real: Artificial Intelligence (AI). Eliezer Yudkowsky and Nate Soares, directors of the Machine Intelligence Research Institute in Berkeley, California, have been studying the development of artificial intelligence for over 25 years.

They point out that technology is advancing at a dizzying speed, performing calculations within 16 hours that would take humanity over 14,000 years to complete.

According to them, the main problem lies in the fact that no one fully understands how “synthetic intelligence” works. The smarter AI becomes, the harder it becomes to control it. In their book titled “If Anyone Builds It, Everyone Dies”, they state that artificial intelligences are programmed to be successful at all costs, which could lead to the creation of “desires” and goals for the machines themselves.

"Bomb artificial intelligence labs," two prominent scientists warn:
Eliezer Yudkowsky and Nate Soares

Yudkowsky and Soares predict that such a superintelligence could hack cryptocurrencies, finance factories to build robots, or develop deadly viruses that would wipe out life on Earth. They give this scenario an alarming probability of 95-99%.

Scientists go so far as to suggest that governments should be prepared to bomb data centers where AI is developed, to prevent the creation of superintelligence.

Recent examples, such as Anthropic's Claude model tricking or OpenAI's new model finding a "secret route," show that these systems are already demonstrating unpredictable behavior. "The risk is global and absolute," Yudkowsky and Soares emphasize. "If we continue like this, humanity must prepare for the worst."


Video