The topic of advanced artificial intelligence (AI) in the future usually carries with it certain fears - for example, that it will be capable of self-learning and become so developed that people will no longer be able to understand it. In the wrong hands, and even on its own, such an advanced algorithm can take over governments and armies, introduce Orwellian methods of surveillance and manipulation, establish control over society, and even control an army of autonomous combat drones on the battlefield.
But some AI experts don't see these fears as valid. In fact, a highly advanced AI may be better able to govern the world than humans. Such fears are themselves dangerous because they can prevent us from realizing such an opportunity.
“Maybe the lack of AI is a danger to humanity,” said Facebook AI researcher Thomas Mikolov at a conference in Prague on the challenges of creating AI at the level of human intelligence.
Mikolov explains that humans, as a species, are very weak in making long-term decisions to bring beneficial results. Humans have cleared rainforests to extract raw materials, not realizing (or not interested) that by doing so they have contributed to the degradation of our planet - perhaps irreversibly.
But an advanced AI system could protect humanity from its own shortsightedness.
“People are poorly able to foresee the distant future, say, 20-30 years ahead. Perhaps by creating an AI that is much smarter than we are, we can help ourselves avoid future disaster."
Of course, Mikolov belongs to a minority of scientists who believe that a highly developed AI will be benevolent. During the conference, many speakers expressed widespread fear of AI, more often in connection with its possible use for dangerous purposes by malicious actors, the danger of which should not be underestimated.
We don't know for sure whether it will be possible to create a general-purpose AI capable of performing any thought task available to humans, and perhaps even better than them.
Promotional video:
The future of advanced AI looks promising, but it raises many ethical concerns. We don't even know all the questions that will have to be answered.
Yet most of the participants in the Prague conference agreed that the rules must be worked out before they are needed. Already, it is necessary to create international treaties, ethics committees, and regulatory bodies at governments, companies, and scientific communities. Implementing such institutions and protocols will reduce the likelihood that a hostile government, an unwary researcher, or even a truly insane "scientist" will trigger malicious AI or use a complex algorithm in another way as a weapon. And in the worst case, we will have remedies.
By introducing the necessary rules and precautions, we will be able to live in harmony with highly developed AI in the future; maybe one day he will even save us.
Vadim Tarabarko