Opinion: Experts are ‘spooked’ by their own achievements in AI

Experts in artificial intelligence are ‘spooked’ by their achievements, comparing the development of the technology to the creation of the atomic bomb. This was stated by Stuart Russell, a professor at the University of California, Berkeley, according to The Guardian.

According to him, the AI community has not yet adapted to the fact that it exerts a large influence on the real world. Previously many studies did not go beyond laboratories, but in recent years the situation has radically changed, he added.

“Now we need to grow up very quickly to catch up,” the professor says.

Russell pointed to a large gap between current technologies and artificial intelligence depicted in films such as Ex Machina. However, the future with machines that are smarter than humans, in his words, is not far away:

“The figures range from 10 years for the most optimistic forecast to several hundred years. But nearly all researchers agree that this will happen in the 21st century”.

The professor emphasized that a machine does not necessarily have to be smarter than humans in every respect to pose a serious risk:

“This is happening right now. Today, algorithms in social networks already have enormous control over our cognitive input”.

In his view, as a result, algorithms manipulate users, brainwashing them for more predictable behavior.

Russell also believes that using AI for military purposes, such as autonomous anti-personnel weapons, raises particular concerns.

“They are very easy to scale, meaning you could fit a million of such trucks into one, open the rear, they would go and destroy an entire city,” Russell said.

According to the expert, the future of AI lies in developing machines that know the true goal is uncertain and must be aligned with humans. However, implementing such an idea is difficult because different people have different preferences, and sometimes they even conflict with each other.

To keep control of the technology in human hands, the scientist called on the international community to conclude treaties and adopt ethical codes regulating it.

In 1995, Russell, co-authored with Peter Norvig, wrote the book “Artificial Intelligence: A Modern Approach”, which became the foundational textbook in introductory university courses in the field.

As a reminder, in September the UN High Commissioner for Human Rights Michelle Bachelet urged to ban certain types of artificial intelligence.

In May, Nobel Prize-winning economist Daniel Kahneman stated that AI will surpass humans, but not soon.

In April, machine learning pioneer Michael I. Jordan urged not to call everything artificial intelligence.

Subscribe to ForkLog News on Telegram: ForkLog AI — all AI news!