Critical Evaluation on Artificial Intelligence: Risks, Uncertainties and Sociological Dimension

On the last three in a row, I wrote comments discussing the financial risks brought by AI technology and concerns about the survival of humanity. This time, I will try to evaluate these comments in general and think about the sociological dimension of AI.

NEITHER DENIAL NOR PANIC

Nobel Prize-winning Prof., also known as the father of artificial intelligence. "People can see concrete threats, such as a nuclear explosion, but they have difficulty grasping abstract threats, such as the human-threatening features of AI. Even I have difficulty grasping risks on an emotional level," says Geoffrey Hinton. Indeed, AI is very different from past technologies. The nuclear bomb is a tool of mass destruction, but it cannot improve itself; only humans can make new bombs and decide to use them. AI has the potential to improve itself and create new AI models; Considering that the more difficult the problems presented to us, the faster they develop, this potential is very close to becoming a reality. At that point, a huge intelligence explosion is expected. Secondly, as AI's infinitely replicable "agent" models developed, the learning speed also increased. The neuronal connections in humans' brains still far exceed those of the most advanced AI model. However, AI has a rapidly growing advantage in the field of learning and sharing. The process of learning and sharing what one learns by speaking and writing is very slow. AI agents can instantly share what they learn among themselves. One can read “Das Capital” in a few minutes and instantly share what they read with other agents. It should not be forgotten that AI can achieve this in 180+ languages.

It is very difficult to think clearly about the disasters that AI may create for humanity, but the notes on the subject of FT writer Soumaya Keynes, the 3rd generation nephew of the famous economist Keynes, may be helpful. I will summarize it.

Some people are too comfortable believing that the development of machines can be easily stopped, while others believe that it is inevitable that AI will destroy humanity. Soumaya argues that, first of all, the concept of "disaster" should be clearly defined. Millions of people may die, but this does not mean the extinction of humanity. There is also the possibility that those left behind will build a better society over time. Second, cognitive biases can often and dramatically make threats appear much larger than they actually are. Conversely, thinking that you will not be personally harmed may also underestimate the danger. AI is not a gamble whose risk can be assessed with exact probabilities. The rules, probabilities and possible scenarios are still unclear. Therefore, it is necessary to accept uncertainty, question prejudices, and improve the preparedness of institutions.

UNCERTAINTIES AND POSSIBILITIES

Uncertainties and possibilities develop or are shaped around these two approaches and under a very strong influence.

Mustafa Süleyman, CEO of Microsoft AI, defines these two approaches as follows: The first approach sees AI not as a new tool, but as a new stage of evolution. Humanity is just a biological bootloader for this new species. This approach claims that a new type of "silicone creature" is emerging, whose rights and freedoms must be mentioned. Secondly, Süleyman puts himself on this side, arguing as a "speciesist approach" that the existence of AI should be subordinate to the existence of humans, and that anthropomorphism (seeing it as if it were human) is very wrong. Of these approaches, both of which are "accelerator", the first is in favor of releasing the process, the second is in favor of accelerating control efforts rather than stopping it. There is also that powerful effect: AI consumes increasingly more resources to develop. Huge financial resources are being poured into the AI ​​field in the hope of making huge profits. In other words; It seems that the capital relationship will decide which of these two approaches will prevail.

If the first approach prevails, humanity will have to share the planet with a "being" that has a tremendous memory and the capacity to use language, calculate, learn and organize. The theory of evolution and human history tell us that species with low adaptation - competition - capacity became extinct over time. There is no power yet to take AI from the hands of capital.