Prof. Petros P. Groumpos, Emeritus Professor, University of Patras
Artificial Intelligence (AI) represents a major technological development, offering enormous potential but also posing serious risks. Today’s systems remain primarily forms of “Narrow AI,” while Superintelligence—recently proposed by American scientists and immediately embraced by President Trump through an executive order—remains, at present, a hypothetical prospect. The possible surpassing of human capabilities raises critical issues concerning safety, control, the economy, defense, and the international balance of power. For this reason, scientific research, international cooperation, institutional mechanisms, and effective safeguards are required. Technological progress must serve humanity, democratic values, and the collective interest, with Cybernetics—a scientific field that predates AI—having the potential to make a substantial contribution to the control of AI systems. In this way, we may ultimately create a useful and harmonious form of AI for the benefit of humanity.
MAIN TEXT
Artificial Intelligence has justifiably emerged as one of the most important technological developments of the 20th and 21st centuries. We expect it to make work more efficient, contribute to medicine, accelerate scientific research, strengthen education and productivity, and provide digital tools capable of writing, analyzing, programming, designing, and carrying out complex tasks with limited human intervention.
However, the history of AI is also characterized by cycles of excessive expectations and disappointments, the so-called “AI Winters.” A characteristic example is Herbert Simon’s prediction in 1965 that, within twenty years, machines would be capable of performing any task that a human could perform. Full Artificial General Intelligence (AGI), however, has not yet been achieved. Today’s systems, such as ChatGPT, Claude, and Gemini, remain primarily forms of “Narrow Artificial Intelligence,” designed to perform specific categories of tasks.
There are also several scientists, as well as other individuals whose views command considerable public attention, who have long warned about the dangers associated with today’s AI. I believe that everyone is following the statements made by former senior executives of major AI companies, as well as by Bill Gates, Elon Musk, Noam Chomsky, and others, and there is no need for me to dwell on these or other such statements.
- The question that now arises is: How far can Artificial Intelligence go, and what will the consequences be if it significantly surpasses current human capabilities?
In recent years, a major development has been unfolding in science and technology: the rapid advancement of Artificial Intelligence (AI) and the growing discussion about the possibility of transitioning to Superintelligence (SI), or “Superhuman Intelligence.”
This discussion has also acquired a political and geopolitical dimension. On September 29, 2026, U.S. President Donald Trump signed an executive order entitled “Inaugurating the Era of Super Intelligence,” marking, according to the U.S. administration, a new phase in the United States’ approach to advanced intelligence technologies.
Superintelligence, however, is not simply another name for today’s AI. It refers to a hypothetical level of artificial intelligence at which a system could significantly surpass human cognitive abilities across a very broad range of activities, including scientific research, technological development, strategic thinking, and the resolution of complex problems.
Many experts emphasize that this concept must be clearly distinguished from today’s AI systems. The use of such an ambitious term for technologies that are still under development may create expectations that do not correspond to their actual capabilities.
If, however, genuine “Superintelligence” ever becomes feasible, its consequences will not be confined to science and the economy. They could affect defense, security, international politics, economic power, and the global balance among states. A state or organization possessing such capabilities would simultaneously acquire a significant advantage and could attain unprecedented technological, economic, and strategic power.
The challenge, therefore, is not to halt scientific and technological progress. It is to ensure that such progress remains beneficial to humanity and that the development of extremely powerful systems takes place on the basis of adequate scientific knowledge, social dialogue, and international safeguards.
It is particularly important—and perhaps one of the most interesting developments of our time—that an increasing number of researchers, scientists, and professionals working in Artificial Intelligence, and doing so over increasingly longer periods of time, recognize that technological progress alone is not sufficient. They believe that the development of sustainable, human-centered, useful, and creative AI requires better education, a deeper understanding of both its capabilities and its risks, as well as strict and effective controls at every stage of its development and use.
Perhaps, therefore, the greatest challenge is not merely to create increasingly intelligent machines, but to establish the conditions under which this intelligence will serve humanity and society. And why not, provided that it eventually becomes technologically feasible, even move toward a controlled Superintelligence that operates safely and responsibly, in harmony with human values, democratic institutions, and the collective interest.
- If “Superintelligence” is ultimately developed according to the same patterns by which AI has evolved throughout its 71-year history, we can be certain that it will fail to achieve its objectives. If, moreover, it comes to be controlled by a single great power, the world may find itself living “in a world” that no one would even dare to imagine.
I therefore believe that the “problems” that will arise from both Artificial Intelligence and “Superintelligence” can be addressed only through the establishment of appropriate safeguards and controls. These controls can, to a large extent, be designed and developed on the basis of the principles and theories of “Cybernetics.”
This, however, is a particularly important and complex subject, which I will explore in greater detail in a forthcoming article. In this way, our leaps will not be leaps into the unknown, but steps toward a better world—a world that we will have the ability to shape with justice, knowledge, wisdom, and responsibility.
Σχόλια