Will the development of the robot's intelligence entail danger? Can he make a strategically correct, but not humane decision, cat. will negatively affect K.-L. sphere of life?

Will the development of the robot's intelligence entail danger? Can he make a strategically correct, but not humane decision, cat. will negatively affect K.-L. sphere of life?

Christopher D Manning: A Neural Network Model That Can Reason (ICLR 2018 invited talk)

Previous questionWhere does the sun get energy for its functioning? If it burns for millions of years, then there must be a source of replenishment of resources?
Next questionIf I get myself a robot like SpotMini, and I order to run to Auchan for a beer, they will sell it, because he is not yet 18 years old?

answers (2)

Answer 1
August, 2021

Almost any decision has negative consequences, especially the “strategically correct” one.

The concept of humanity is so confused that almost any decision contradicts it in one form or another.

Robot with intelligence, by definition, self-learning. This means identifying cause-and-effect relationships, which is constantly performed by humans with errors, and by robots as well. And this, of course, can be dangerous, for example, in medicine.

The correct formulation of the problem is also a big problem. We are used to solving narrow problems in a certain "ecological niche" when many things are "obvious". But for artificial intelligence, everything that has not been trained is not obvious. He immediately finds a much more optimal solution, like Elektronik, who scores the ball into the nearest goal.

It only saves that "robots", for purely practical reasons, are introduced as highly specialized systems and often with parallel control of people: vacuum cleaners, navigators, homing missiles, recommendation systems. Liability is insured, and problems are gradually being solved.

In addition, for practical reasons, it is much easier to hire a person for a role with the requirements of purely human qualities (the same questions of humanism). A great example is a Chinese company that recognizes captchas for web bots (pictures proving that the user is not a robot) used a thousand Chinese people to do this. I didn’t check if this was really so, but I saw an advertisement and it perfectly illustrates the possible division of labor.

So it will be like with any other automation: the most routine and unusual functions will be gradually transferred to computers under supervision people. What computers cannot actually handle will be passed back to people.

PS. By the way, I think that not robots will be humanized, but people (as a society) will be rationalized.

Answer 2
August, 2021

Robots are algorithms. If their software is explicitly required not to harm living beings, he will not be able to make a decision that contradicts this requirement.

So, I think, if an uprising of machines is possible, it is only provoked by a human factor, that is, by the intervention of a specific person in this process (for example, the ban on harm to living beings will be lifted, or killer robots will be created, etc.). etc.).

Related question

How do you feel about the rapid development of artificial intelligence? Do you really think he will destroy humanity?

Read more

Why, when you bring a cat to the mirror, he looks anywhere, but not at himself?

Read more

Should we impose a moratorium on the development of robots and artificial intelligence in order to avoid total unemployment?

Read more

Public Hearing on Section 232 Investigation of Steel Imports on National Security

If in the near future 70-80% of families in developed countries will be same-sex, how will this affect the future of humanity?

Read more

A question timed to coincide with the discovery of new planets: "What race will humanity be in relation to other intelligent life forms?"

Read more

How can a graduate student get a job if there is no work experience, but, at the same time, experience is required everywhere? (sphere of economy, finance)?

Read more

A paranoid question, but can photos of a person that he did not upload / send to anyone can get into the network?

Read more

Why is science innovation not going into mass production? Why is there news about the achievements of science, but in life you will not meet them?

Read more

Mad Scientist Conference: Learning in 2050 | Thursday, August 9

Is it possible for a hollow space body to exist three orders of magnitude larger than an asteroid? Is there a scenario for its origin and the development of intelligent life within it?

Read more

It is becoming more and more dangerous to express your thoughts on the Internet. They can make fun of the whole world, dismiss, arrest. Is there a chance that a massive exodus from social networks, etc., will begin soon?

Read more

How to prove to all sorts of old men that games are not just entertainment, but also a cultural feature and the development of technology?

Read more

Why does the contact rail in the subway run along the edge of the platform, and not opposite? Indeed, when the path falls on a person, he represents a serious danger.

Read more

How can I not pay attention to insults on the Internet - although I know it's just a bunch of letters, but I still can't calm down?

Read more

Things Mr. Welch is No Longer Allowed to do in a RPG #1-2450 Reading Compilation

If a citizen of the Russian Federation posts on a social network a post that is criminally punishable in the Russian Federation, but from the territory of another country, can he be prosecuted for this under the laws of the Russian Federation?

Read more

I will write very abbreviatedly, 18 years old, closed. They hounded at school, now I am studying at the university. I want to make friends, but a lot of people just don't fit for me. How can you find someone like you?

Read more