It is we who are dangerous, not the machines

ARTIFICIAL INTELLIGENCE: A computer can beat you, but strive to see the difference between cats and dogs. Does that mean it's not intelligent?

Literary critic in MODERN TIMES.
Published: 2020-02-07
Artificial Intelligence - A Guide for Thinking Humans
Author: Melanie mitchell
Pelican Books, United Kingdom

Experts on Artificial Intelligence (KI) is mainly divided into two camps: Those who feel confident that machines will pass humans in intelligence in a relatively short time and who fear what consequences this will have for humanity, and those who feel confident that machines can never become more intelligent than us and therefore believe there is little to fear.

Author Melanie Mitchell definitely belongs to the latter category.

Mitchell is a professor of informatics at the University of Portland and has worked with analog thinking, complex systems, genetic algorithms and cellular automata (mathematical modeling). She has several books and publications behind her, among other things An Introduction to Genetic Algorithms (1996). She claims that even though a machine has managed to make itself unbeatable in chess, it does not suffice for anything else.

I think she seems too laid back in her views. It is true that what is easy to learn for humans is difficult for machines, but the opposite also applies. The biggest problem, on the other hand, is that machines and humans have trouble understanding each other. It's hard for us to understand why intelligent machines strive for something that even small children can handle as easily as nothing - such as distinguishing between a dog and a cat in a photo.

We humans tend to overestimate the intelligence of machines and to underestimate them
our own.

Computers are lightning fast to learn the rules of different games, but have great challenges in explaining why they do as they do in a way that we can understand. It's hardly understandable to the machine either. But when we look at how fast computer science has evolved, little can be excluded in the future. The book is far too small about the fusion between man and machine.

Subscription half year NOK 450

Explanation Problem

A person can easily explain why they chose to walk in the park rather than wash the house. But how can a machine explain why it chose a specific feature on the chessboard and not one of the trillion other options? It can't. Nor does it understand that it has won, or what it means to "win". Nor can it enjoy the victory. We can therefore…

Dear reader. You have already read the 4 free articles of the month. How about supporting MODERN TIMES by drawing a running online Subscription for free access to all articles?

Leave a comment

(We use Akismet to reduce spam.)