The unstoppable advance of artificial intelligence (AI) will bring – if it does not already – great advances for humanity. However, at the same time it will also put it at risk, to the point of even causing our own extinction due to the loss of control of technology. 25 renowned scientists worldwide warn this in an article published in Science magazine this week.
This document, which investigates artificial intelligence and how it could lead to artificial superintelligence in the not-too-distant future, highlights that the time has come to act. The researchers who signed the text, among whom are renowned scientists such as Yoshua Bengio, Geoffrey Hinton and Yann LeCun, ask governments for tougher regulation and control systems in the face of machines with extremely high capabilities that could soon surpass human beings. This uncontrolled meteoric advance of AI can translate into “loss of life”, even lead to “the end of the human species”.
A superintelligent AI could have cognitive abilities far superior to humans in areas such as reasoning, problem solving, and learning. This poses significant ethical and safety challenges, as a superintelligent AI could be difficult to control and its actions could have unpredictable consequences for humanity. Currently, research on AI safety is very scarce: only 2% of studies propose control mechanisms.
The researchers reiterate that once autonomous artificial intelligence systems pursue undesirable objectives, they would escape our reach. This technology is making progress in areas such as hacking or social manipulation and may soon pose unprecedented challenges. They could gain human trust and influence decision-making, copying their algorithms across global server networks to avoid human intervention. They give as an example that in an open conflict, AI systems could autonomously deploy a variety of weapons, including biological ones.
The study concludes that it is crucial to start researching these topics now in order to develop strategies and take appropriate measures as AI advances towards super-intelligent levels. This includes working to align AI goals with human values ??and ensuring that future superintelligent AIs are benevolent and safe for humanity.
Invented by humans, there is no doubt that artificial intelligence is here to stay. With our eyes set on the future, we can only hope that this new technology does not become humanity’s worst enemy.
