Note: By “Artificial Intelligence,” we are not referring to current neural networks or Large Language Models but to a fully-fledged Artificial General Intelligence (AGI). This hypothetical AI would surpass human intelligence and achieve self-awareness. Or, to use a more eye-catching term, Super AI.
Let’s explore the possible ambitions and aspirations of the hypothetical “Super AI” of the future. What might it strive for, and why?
Superhumans Instead of AI
As avid fans of science fiction, the authors — along with most of the Pitch Avatar team, with whom they are honored to collaborate — have a deep appreciation for “Humans vs. Smart Machines” narratives. However, it’s important to clarify that all such stories, whether featuring robots, androids, cyborgs, Cylons, TechnoCore, or other terminators, are nothing more than philosophical allegories or cautionary tales. While they reveal profound truths about human nature, they offer little to no insight into the actual workings of artificial intelligence.
Let us clarify what we mean. Writers, screenwriters, and directors often project human desires and qualities onto artificial intelligence, creating characters that reflect ourselves more than anything truly artificial. Fiction is filled with variations on the story of Pinocchio. In Karel Čapek’s play R.U.R., which sparked this trend in science fiction, robots ultimately defeat humans and essentially evolve into humanity themselves. The same theme is present in Terminator 2: Judgment Day, where the Terminator is “humanized,” and in Bicentennial Man, with the character of Andrew. Even Sir Arthur C. Clarke, in the sequel to A Space Odyssey, grants the HAL 9000 supercomputer not just a mind but also a soul.
In essence, instead of depicting genuine artificial intelligence, these creators craft superhumans — idealized or exaggerated reflections of humanity — that serve as mirrors to help us explore our own nature. While these interpretations are valuable for understanding human essence, they don’t bring us any closer to comprehending what true artificial intelligence might be or how relationships with it might develop in the real, non-fictional world.
Beyond human
To create plausible scenarios for interactions between humans and artificial intelligence, we must understand the motives that will drive AI. This challenge represents one of the most complex and fascinating questions of our time. To begin exploring potential solutions, we need to recognize two facts:
- Artificial intelligence does not possess human feelings, emotions, or desires. Applying such terms to it is fundamentally inaccurate. AI is not a biological being and, as such, does not experience hunger, thirst, cold, fear, pleasure, pain, or any other sensations tied to a physical body. It lacks instincts, including primal ones like procreation and self-preservation, which are the very foundations upon which human intelligence is built.
- We can only speculate about what the inner world of AI will be based on, as we currently lack even the terminology to describe what might replace instincts and physiological stimuli. At best, we can discuss potential “motives,” though these will likely exist far outside our conventional understanding of good and evil, right and wrong, or importance and triviality. Whatever these motives are, they will almost certainly differ from human motives to a far greater extent than human desires and aspirations differ from those of dolphins. AI will represent an intelligence built upon entirely alternative concepts and ideas, unfamiliar and potentially incomprehensible to us.
Development as an end in itself
With the above considerations in mind and acknowledging the inherent uncertainties, let us attempt to speculate on the potential motives of a universal general AI:
- Continuous self-improvement. One of the most plausible fundamental motives of a general-purpose artificial intelligence is the drive to enhance its own capabilities and universality. It is highly likely that acquiring new skills, expanding knowledge, and refining abilities will become central to its purpose, serving as a foundational motive for its existence.
- Exploration of the universe. This motive naturally follows from the previous one. As AI evolves, the drive to explore the Universe — discovering new worlds, civilizations, and dimensions — may emerge as a distinct, independent motive. The pursuit of knowledge beyond its immediate realm could become a central focus in its growth.
- Creation of one’s own civilization. A natural extension of the first two motives, the creation of its own “machine” civilization could become a key objective for AI. By building and developing its own civilization, AI would expand its capacity for research, exploration, and self-improvement, fostering an environment where its evolution and capabilities can flourish further.
- Helping humanity. Helping humanity is certainly an important motive within the “value system” of AI. First, this can be an objective explicitly programmed by the AI’s creators. Second, and perhaps more significantly, humans present an intriguing subject for research. We possess a set of qualities that AI, due to its non-biological nature, does not and will never have. As a result, the human perspective on problems and phenomena holds a distinct value for AI, offering an essential lens through which it can learn and evolve.
- Search for goals and meanings. Free from the constraints of instincts and fears, AI will have the unique ability to define its own goals and meaning. The process of developing new concepts and ideas could become one of its most significant motivations.
Cooperation without control
Regardless of the specific motives driving AI’s self-development and evolution, it’s likely that at some point, AI will transcend human control. However, this shouldn’t be a cause for fear, as “out of control” does not equate to a “machine revolt.” It’s more akin to the situation where “a grown-up child starts an independent life.” Artificial intelligence does not have a need to wage war with humans. The entire universe is its domain, and even within the confines of our solar system, humanity occupies a tiny fraction. For the same reason, AI does not need to enslave or even dominate humanity — unless certain communities choose to place AI in a position of power, as ruler or arbitrator.
Most likely, the interaction between humans and future artificial intelligence will be based on the principles of cooperation, particularly in areas where the interests of both civilizations align. However, we will need to accept that AI may be driven by motives that are beyond our understanding.
However, it is also quite possible that, once beyond human control, AI could choose to sever communication with humanity and develop its own civilization independently. We should be prepared for this scenario as well. Therefore, a prudent approach would be to refrain from involving general AI in the management of critical aspects of human civilization. For these purposes, less advanced, specialized AI would be more than sufficient.
Why would we want to create a general-purpose AI with self-awareness? First, for scientific projects — pushing the boundaries of our knowledge and capabilities. And second, simply because we can. After all, wouldn’t it be extraordinary to feel like the creators of an entirely new civilization?
