What would happen if conscious artificial intelligence took over humanity?

⇧ [VIDÉO] You may also like this partner content (after advertising)

Imagine yourself evolving in a world controlled by conscious robots, where your least actions are valued and notified. It could be an episode of Black Mirror, as a fact in the fairly near future. In this scenario on the border between fantasy and science, which areas of our lives will be affected and in what way? According to a UNESCO text, artificial intelligence can ” Widening the gaps and inequalities that exist in the world If it is not controlled. If conscious AI widely recognized its “slave” role and no longer accepted it, what would happen?

What is consciousness? The answer is not very clear, because ” One of the hardest words to define According to the philosopher André Comte Sponville. It may mean the ability to know one’s truth and judge it (self-criticism), for example to make value judgments about one’s actions. It will therefore be accompanied by morals (“good” or “bad”) and a body.

Last year, neuroscientists identified three types of consciousness, following the different types of information-processing accounts in the brain: recognition of the unchanging subconscious (C0), selection of information to publish globally and thus making it flexibly available for computation and reporting (C1), and self-monitoring. for these calculations, leading to a subjective sense of certainty or error (C2). This last classification—corresponding to a strong artificial intelligence—appears to largely correspond to what one would expect of a “conscious” robot.

According to these researchers, Today’s machines still mainly perform arithmetic operations that reflect unconscious processing (C0) in the human brain “And we are still very far from reaching the C2 level. Many scientists agree that it would be impossible, and that computers would never be able to feel emotions the way we do. It is certainly possible to give machines some form of “feel,” but that remains just a simulation.

Other scientists believe that another form of (artificial) consciousness can be visualized, by studying the structures that allow the human brain to generate consciousness, and then transferring that knowledge to computer algorithms. Even if it seems very far from our present, what will happen to humanity?

Transfer the correct values ​​to the bots

For such a scenario to happen, the human must have already decided on it in the norm. Thus, giving a robot a “good” or “bad” pronoun depends above all on its creator. The danger, for example, is that a machine discriminates against groups of people because it has been programmed to do so. Computer scientist Mo Gawdat wrote in his book that the challenge is to transfer the right values ​​and ethics to robots. ” Artificial intelligence (AI) will take that seed and create a tree that provides an abundance of the same seed. If we use love and compassion [sur les réseaux sociaux par exemple]AI will also use these principles. We are like the parents of an amazing child: one day he will be independent. Our role is to ensure he has the right tools ’ concludes the author.

However, there is a risk of drifting and losing control of the machines. One could imagine the revolution of humanoid robots if they recognized their “human” estrangement and tried to turn things around. Marvin Minsky, the American computer scientist who co-created the first forms of artificial intelligence, has reported to Life magazine In 1970: Once the computers take over, there is no going back. We will only live because they will. We can consider ourselves lucky if they keep us as pets “.

“As the capabilities of AI-based technologies expand, the potential for them to be exploited for criminal purposes increases.”

Depending on societal choices, AI can become a weapon against individual liberties and serve social control. The effects will be enormous on all aspects of society: job replacement, education control, environmental degradation, etc. “Bad” AI can lead to misinformation and even increase crime. Here’s what Lewis Griffin, a computer science researcher at University College London, thinks: As the capabilities of AI-based technologies expand, the potential for them to be exploited for criminal purposes increases “.

The English researcher and his team compiled a list of twenty potentially AI-related crimes and ranked them in order of anxiety (low, medium, high) and in terms of victim harm, criminal profit, criminal achievement potential, and difficulty quitting smoking. The workshop, titled “Artificial Intelligence and the Crime of the Future,” brought together representatives from academia, police, defence, government and the private sector.

General classification of violations resulting from the workshop. For each crime, the colored bars indicate the average ranking for the four dimensions: harm to victims (in yellow), criminal profit (green), possibility of criminal investigation (red), and difficulty of arrest (blue). Bars above (or below) the line indicate that crime is more (or less) a concern in this dimension. Error bars indicate the interquartile range. The offenses listed in the same column should be considered of equal importance. Anxiety increases with the column from left to right. © Caldwell, Andrews et al. 2020

Of the six most threatening categories, five have broad societal impact, such as misusing autonomous vehicle technology for terrorist attacks or causing accidents, and those involving fake AI-generated content. In this case, the false information can violate the identity of the person to request private access, or even damage the reputation of a known person. The ” deep fake They are difficult to detect and combat and therefore rather dangerous. Also, fake content written by AI would definitely confuse human minds to distinguish the real from the fake.

The researchers then highlighted “medium severity” threats, such as financial market manipulation, cyber-attacks, data corruption, and fraud or gun control for criminal purposes. Although intimidating, this last threat is rated as (medium) because it is difficult to implement, and military equipment is well protected.

For this same reason, “robot thieves” rank among the least dangerous threats, as they can be easily stopped. In addition, imitation may consist of the manufacture and sale of pseudo-cultural content (music, paintings, etc.), but seriousness is not considered important.

Ethical issues for a fairly near future

The main danger of conscious AI is to get out of our moral and legal framework, and lose control of it. Closer to our present, the first UNESCO universal text covering all areas related to artificial intelligence, its benefits and risks to society, was adopted on November 24, 2021.

The Recommendation describes the values ​​and principles that should guide policy and legal actions in the development of AI. Among them are respect for human rights and inclusion (non-discrimination, gender equality, etc.); Contribute to the sustainable development of research and use of artificial intelligence; AI Integrity (risk assessment, data protection, prohibition of the use of AI for purposes of social classification or mass surveillance).

Breakthroughs in algorithms represented by cognitive computing are driving the continuous penetration of AI into areas such as education, commerce and medical treatment, to create space for AI services. Chinese researchers write. As for the human interest in who controls who among humanity and intelligent machines, the answer is that AI can only become a service provider to humans, demonstrating the rationality of the value of AI. “.

While most scientists agree with this, others worry about the scale that AI development could get. ” Already today, AI systems detect when a human is trying to modify their behavior and sometimes do everything to reject and circumvent this interference if it conflicts with the initial goal of AI. “, Warn weather Rachid El Karaoui, Director of the Distributed Programming Lab at EPFL (Switzerland). ” You have to act subtle and fast so that the AI ​​believes it is making all the decisions itself. Then erase the effects of human intervention So beware, even if the machine takes over man not for tomorrow.

Leave a Reply

Your email address will not be published.