INSURRECTION/ SP MPs rebel, drag Gjati to vote on territorial division!

 Kliko këtu për të tjera
Voxpress:

Could Artificial Intelligence kill us all? And how exactly?

2026-09-21 20:38:05Kosova&Bota SHKRUAR NGA REDAKSIA VOX
Could Artificial Intelligence kill us all? And how exactly?
Artificial intelligence researchers say this technology could wipe out humanity if it achieves "superintelligence"

The debate over whether artificial intelligence could threaten the very survival of humanity has reignited in recent days, following public warnings from AI professionals that the development of increasingly powerful systems is bringing us dangerously close to destruction.

The reason for this was the decision of Jacob Coxon, an employee of the artificial intelligence company Anthropic, to leave the company, arguing that the direction of the research was too dangerous. A little later, his colleague Evan Humpinger estimated that the probability that advanced AI would lead to the extinction of the human species within the next decade exceeds 10%.

As CNN reports in an analysis, these claims have provoked strong reactions from politicians, technology circles and public figures, highlighting a fundamental question: if artificial intelligence could truly wipe out humanity, how exactly would that happen?

The most pessimistic experts talk about scenarios involving biological weapons, autonomous robots or the taking over of critical infrastructure. Other scientists, however, argue that these predictions are based on assumptions about technologies that do not yet exist and that there are no specific factors that would make such a sequence of events possible.

The threat of an artificial pandemic

One of the most discussed scenarios involves biology. Some researchers have suggested that highly advanced artificial intelligence could help create a dangerous pathogen or use humans as "go-betweens" for the development of biological weapons.

Thomas Larsen, a researcher at the AI ​​Futures Project and a former member of the Machine Intelligence Research Institute, argues that a system far more sophisticated than today’s models could manipulate humans into performing experiments on its behalf. This concern is exacerbated by the fact that researchers are already using AI models for scientific work, including in areas related to biology and pharmaceutical research.

However, many scientists argue that moving from theoretical knowledge to a real, operational, and highly dangerous biological weapon is a huge step. Eric Singh, a professor of machine learning at Carnegie Mellon University who holds a doctorate in both computer science and microbiology, likens the process to trying to build Lego without instructions.

“It’s not like you throw all the pieces on the ground and they just come together,” he explains, noting that biology requires extreme precision, special conditions and complex processes. At the same time, he notes that there are already strict control mechanisms around laboratories, materials and supply chains, which act as important barriers to the development of such threats.

The scenario of autonomous robots

A second scenario considered by supporters of the "existential threat" theory involves the creation of autonomous robots that can reproduce and operate without human intervention.

Nate Soares, president of the Machine Intelligence Research Institute, argues that the tipping point would be when an AI could create the infrastructure it needs on its own: factories, energy systems, and new cars. “When you have robots that can build the infrastructure to produce more robots, then you’re talking about a new mechanical life form,” he argues.

However, this scenario is far from the reality of today's technology. Despite the huge investments of tycoons like Elon Musk in humanoid robots, these systems are still subject to serious limitations and lack the ability to produce and expand autonomously.

Can AI control nuclear weapons?

Another question concerns nuclear weapons and whether an advanced artificial intelligence could gain access to them.

Heidi Klaaf, a senior scientist at the AI ​​Now Institute and former security engineer at OpenAI, points out that nuclear systems are designed with extremely strict security protocols and are largely isolated from the internet. Even the Stuxnet cyber virus, which damaged Iran's nuclear facilities, required physical access to the system via a USB, he says.

Herbert Lin, a researcher and security expert at Stanford University, argues that the real threat still lies with nuclear weapons themselves and human decisions about them, not with a hypothetical scenario of AI taking control of the systems.

The "self-improvement" that scares experts

For some researchers, the real threat is not necessarily today’s artificial intelligence, but the possibility of a so-called “artificial superintelligence.” The theory goes that if a system gains the ability to improve its capabilities on its own, its evolution could become so rapid that humans can no longer control it.

Soares argues that the problem is not an AI that has “hostile” intentions, but a system that pursues a goal without understanding human value. “If a system wants more computing resources to achieve its goal and has no reason to protect people, then it can transform the world into something in which we will not be able to survive,” he argues.

Between fear and hyperbole

The debate over the future of artificial intelligence is not just about the technology, but also about how people perceive its power and limits. Thomas Larsen tried to give more concrete form to possible doomsday scenarios through his analyses of “AI 2027” and “AI 2040,” examining different paths for the development and regulation of artificial intelligence. However, even in his most optimistic scenario (where there is a global agreement on how AI should develop and humanity has already begun to expand into space) the result remains, according to him, the dominance of mechanical systems at some unspecified point in the future.

He states that there is no doubt that the path towards superintelligence will continue, unless conscious decisions are made to change the direction of development.

Herbert Lin, on the other hand, believes that much of the distance between supporters of doomsday scenarios and skeptics arises from the different ways in which people experience technological development. As he explains, there is something particularly exciting about the experience of creating a program that seems to “come alive” and respond to one’s commands. It is precisely this sense of unprecedented discovery, according to Lin, that may lead some researchers to project today’s impressive progress in artificial intelligence into a future without limits, even in extreme scenarios where technological development escapes all human control.


Në përputhje me standarded JTI

Video

SHIKO KANALIN NE YOUTUBE
×