Dario Amodei, CEO of one of the world's leading companies for artificial intelligence Anthropic, warns humanity that it is on the verge of being granted "almost unimaginable power" for which it is "completely unprepared".
In the extensive text entitled "Adolescence technologies", Amodei analyzes in detail the potential risks of "powerful", generative artificial intelligence. It addresses security and military threats, possible political and economic disruptions, and labor market transformation. The author additionally considers hypothetical scenarios of the emergence of synthetic entities, with potentially existential consequences.
Amodei strives to provide the reader with a comprehensive overview of the possible risks of these revolutionary technologies on society, the economy and the very nature of life.
AI as an amplifier of human intentions
This expert warns that we are facing a period in which humanity creates technology that, on the one hand, exponentially increases human abilities, but at the same time erases the protective barriers that until now protected us from our own destructiveness. Powerful artificial intelligence will not just be a smarter tool, but a universal "intention amplifier" - good and bad - giving individuals capacities previously reserved for states, institutions and narrow circles of top experts.
Amodei warns that while today a "disturbed loner" may commit a local act of violence, he is unable to create weapon for mass destruction, powerful AI changes that. She could guide him, step by step, through the process of designing, synthesizing, and disseminating a potentially lethal biological agent, practically elevating the average individual to the level of an expert with a Ph.D.
At the same time, the text draws attention to extreme, but not impossible, scenarios of the creation of the so-called mirror life organisms - synthetic forms of life whose molecules are arranged "reversely" in relation to natural life, which makes them biologically incompatible with existing organisms. Their impact on the environment could be devastating: they could spread uncontrollably and, in the worst case, threaten all ecosystems on Earth.
Global risks
All of the above risks are further amplified by the fact that artificial intelligence is already accelerating its own development – writing code, designing, experimenting and building the next generation of systems – while society and regulation lag behind or stagnate.
In terms of security, things can be even more dangerous: if one country gains a decisive advantage in the development of artificial intelligence and puts under control what the author calls "the land of data center geniuses" - powerful AI systems that operate behind the screen - within its own military apparatus, other countries could become virtually unable to defend themselves. They would be outsmarted and outgunned at every turn, like a "war between men and mice." This situation, according to the CEO of Anthropic, opens the possibility of the emergence of the first real global, totalitarian dictatorship.
In addition to the inherent security risks that AI can pose on its own or through misuse by individuals, organizations or states, there is also a serious economic challenge. In the future, artificial intelligence will quickly surpass the cognitive abilities of humans and will replace a large number of jobs in a short period of time, which can lead to a potential destabilization of the workforce. The speed of development, the ability of AI systems to compensate for shortcomings "on the fly", the wide range of tasks they can undertake and the concentration of power in the hands of a small number of actors can cause market disruptions, growing unemployment and inequality, the formation of new lower economic layers and the accumulation of even more wealth among individuals and a small number of states.
Maturity test
That is why the next period can be considered a test of the maturity of civilization: society will have to find ways to preemptively restrain and collectively control this power, otherwise it could be faced with consequences whose scope exceeds the capacities and possibilities for their correction, concludes Dario Amodei.
