The most serious problem with the application of AI is the loss of human control over decision-making. That is, the delegation of decision-making authority to AI. And this process will not be violent or imposed from the outside by some Big Brother. People will voluntarily surrender control because it will be easier, faster, and more reliable that way. Just as today almost no one multiplies or divides by hand in columns anymore, but uses calculators; just as almost no one memorizes a route from a paper map, preferring to follow a navigator’s instructions.
People already trust many “smart” devices with part of their decisions — blood pressure monitors, microscopes, washing machines, electricity generators, aircraft navigation systems, and many others.
AI is simply the next stage of the technological process. And the technological process throughout history has served one main purpose: to free humans from hard and tedious labor. That’s how the Industrial Revolution destroyed artisanal production. By the way, it also made slave labor economically meaningless.
But there is one fundamental difference. A significant portion of human decisions arise at the level of emotions, shaped by ancient brain structures, and logical explanations are fitted afterward. These emotional reactions are biochemical algorithms honed by millions of years of evolution. Until now, no technology has understood a person better than they understand themselves.
With the advent of AI, however, this could change dramatically. Once AI gains access to brain signals, it will be able to analyze the biochemistry of emotions, optimize it, and offer a person decisions that will objectively be better. Exactly how this will happen, I naturally don’t know. Perhaps it will help choose a more suitable partner or profession. And if people become convinced that such recommendations, on average, yield better results than their own intuitive choices or the advice of parents and mentors, the shift toward choosing AI will become widespread.
And that’s where the risk appears. Just as calculators have unlearned people how to do mental math, and spell-checkers in text editors have unlearned people how to write without mistakes, delegating most decisions to AI will make humans extremely vulnerable in the event of system failures, accidents, or — who knows — possible malicious intentions on the part of the AI.
The only thing holding us back from this future for now is energy constraints. Every new leap in AI development requires enormous amounts of electricity that current power capacities cannot provide. Therefore, in the next ten to twenty years, a complete surrender of control is unlikely.
However, right now leading scientific centers and major corporations are actively working on controlled hot thermonuclear fusion and cold nuclear fusion. According to optimistic forecasts, the first realistic results are expected in the 2040s–2050s. When fusion energy provides practically unlimited resources, humanity will gain the ability to create a new level of AI — and it is precisely then that the problem of losing control over our own decisions will arise in full force.
