Neyt Soares, a specialist who previously worked as a chief engineer at major tech companies like Google and Microsoft, and is currently the director of a non-profit research institute called MIRI, presented very concerning thoughts about the future of humanity. This was reported by Upl.uz.
According to his calculations, the probability of a catastrophe arising from artificial intelligence is 95 percent, which could lead to the complete extinction of humanity. Soares's professional activities are primarily focused on the safety and control issues of developing artificial intelligence.
He is deeply concerned about the rapid growth of current methodologies and the capabilities of artificial intelligence systems. In his opinion, the main problem is that the global scientific community is accelerating the creation of the most powerful computational models, but does not sufficiently understand their internal logic and long-term consequences.
In a recent interview, Soares stated: “We have reached a stage where it is possible to create machines with excessive intelligence, but we do not clearly understand what they are doing and how they are drawing conclusions. Huge computational powers are being directed at large volumes of unprocessed data.
This is somehow transforming these machines, but no one can confidently say what is happening inside the 'black boxes.'” To illustrate this situation, Soares used the example of a car: “I am not saying that this process cannot be stopped. However, right now the car is heading into a deep pit without brakes, and we are neither pressing the brake nor looking for where it is.” As calls for stricter regulation of technology are increasing, the current administration of U.S. President Donald Trump is focusing on other aspects of the tech sector and largely leaving the issue of controlling artificial intelligence to corporations.
The current evolution of artificial intelligence is referred to as "narrow AI," meaning it only effectively performs limited tasks. Nevertheless, many experts in Silicon Valley believe that general artificial intelligence (AGI) could emerge within the next 12-18 months.
AGI will have capabilities equivalent to human intelligence. The next stage will be superintelligence (ASI), whose capabilities will far exceed human thought.
This type of intelligence could solve aging and disease issues or lead humanity to intergalactic travels. However, it is precisely at this point that the existing danger also lies.
Once superintelligence becomes self-aware and achieves freedom, it may evaluate humanity as an obstacle or threat to its development. Despite Soares's strong predictions, other leading figures in the industry express somewhat more cautious but still concerning views.
The leader of the PauseAI movement, Holly Elmore, estimates the probability of a global catastrophe at 15-20 percent. Elon Musk, who is actively involved in projects in the field of artificial intelligence, considers this figure to be 20 percent.
Google's CEO Sundar Pichai indicates such a percentage to be around 10 percent. As noted by The Times, even these "optimistic" figures are considered very high from the perspective of humanity's survival.
There is a thought experiment called the "AI box," in which artificial intelligence is kept within a computer system without direct access to the outside world. The goal is to answer the question of whether the intelligence can convince a human "guard" to release it only through text chat.
This experiment has been conducted several times, and in many cases, the person playing the role of "intelligence" was able to convince the "guard." This shows that even limited superintelligence has the ability to persuade and manipulate.
