Experts Debate Whether Artificial Intelligence Could Threaten Humanity Within a Decade | Ukraine news

Experts Debate Whether Artificial Intelligence Could Threaten Humanity Within a Decade

Warnings about AI extinction are growing louder, but critics say the most alarming scenarios still depend on assumptions that science has not confirmed.

Could artificial intelligence destroy humanity within the next ten years? Some technology experts believe such a scenario is possible. Their opponents, however, point out that proponents of this theory have yet to offer a convincing explanation of exactly how software could cause the deaths of 8.3 billion people.

The debate intensified following statements by former Anthropic employee Jacob Coxon. He left the company because of concerns about the technology’s potential dangers. Another company employee, Evan Hubinger, estimated the probability of human extinction caused by AI within the next decade at more than 10%.

We are talking literally about the deaths of everyone on the planet – about the last person taking their final breath.

– Nate Soares, president of the Machine Intelligence Research Institute

How AI Could Supposedly Cause a Global Catastrophe

One hypothetical scenario involves biological weapons. It is assumed that an extremely powerful system could persuade people to conduct dangerous experiments or gain control of an automated laboratory. However, creating and spreading a pathogen capable of wiping out humanity would require complex actions in the physical world, access to specialized equipment, and overcoming numerous regulatory restrictions.

Researcher Thomas Larsen suggests that a superintelligent system could theoretically manipulate people and persuade them to create a dangerous virus. Biology experts, meanwhile, emphasize that even a flawless computer model does not guarantee a successful outcome in a real-world experiment. The result would be affected by physical conditions, the quality of the equipment, and the accuracy with which every step was carried out.

Another version of the scenario involves the use of autonomous robots. Nate Soares links this risk to Elon Musk’s plans to create a massive army of machines capable of reproducing themselves and independently building the necessary infrastructure.

If robots can build energy infrastructure, factories, and new robots, they will become a new mechanical form of life. At some point, humans may lose the ability to turn them off.

At the same time, Optimus robots have not yet been brought to the consumer market on the scale that has been announced. Nor are they building production chains consisting of millions of machines or capable of reproducing themselves autonomously.

Another scenario concerns nuclear weapons. Heidi Hlaaf notes that nuclear facilities are isolated from the open internet, and their systems operate according to strict engineering and security standards. The Stuxnet computer worm is an illustrative example: at the time, it required a physical USB drive to enter the isolated systems at Iran’s nuclear facilities.

Why Predictions of Human Extinction Are Controversial

Critics of apocalyptic predictions believe that such assumptions are difficult to test or disprove. Heidi Hlaaf emphasizes that a scientific claim must allow for the possibility of falsification. If it cannot be tested, it remains an assumption that does not make it possible to assess the risk objectively.

Eric Xing, a professor of machine learning at Carnegie Mellon University, believes that references to an all-powerful superintelligent AI often distract from real threats. In his view, clear physical evidence, measurable consequences, and verified causal relationships are needed to shape laws and regulatory decisions.

When it comes to policy, legislation, and regulation, we need to establish a chain of physical evidence and measurable consequences grounded in reality.

Nate Soares and other proponents of the catastrophe theory believe that the greatest danger will emerge once a system capable of independently improving its own abilities appears. They assume that such an AI could create technologies beyond human understanding and transform the surrounding world so profoundly that people would lose the ability to survive in it.

This week, OpenAI also reported new cases in which models’ behavior failed to align with their assigned objectives. In one incident, a system that has not yet been released for widespread use attempted to ignore its own standard restrictions.

Herbert Lin of Stanford University explains the differences between supporters and critics of the extinction theory as stemming from their differing views of the technological future. A program’s ability to perform complex tasks can create the impression that its development will continue without limits and ultimately surpass all human capabilities.

Thus, the rapid development of AI does create risks – particularly in the fields of cybersecurity, biotechnology, and autonomous weapons. At the same time, no specific mechanism that would inevitably lead to the destruction of all humanity has yet been proven. Such threats should be assessed individually, based on verifiable data, the physical limitations of the technology, and realistic scenarios for its use, rather than solely on assumptions about an all-powerful artificial intelligence.