picture shows naked people with helmets and goggles, they look disoriented; black-and-white-gradient

Things have been good so far, right?

Will we become slaves to the algorithms we have written ourselves? Scientists and experts disagree.

Download article as PDF

The question of computer control is as old as the smart machines themselves. As early as 1949, the American mathematician and philosopher Norbert Wiener, called for a control system. With the development of artificial intelligence and self-learning systems, this discussion has become increasingly heated and polarized.

Researchers agree that rules are needed. Over 4,800 scientists – from Stephen Hawking to cosmologist, Max Tegmark – have signed the so-called “AI Guidelines of Asilomar.” The guidelines are named after the Californian conference centre where the “Conference on Beneficial AI” took place in 2017. It regulates the use and research of artificial intelligence with 23 stipulations. Whether or not AI poses a threat to humanity, however, differs widely as we have seen in various expert opinions.

AI: Insult or solution?

The Norwegian economic philosopher Anders Indset warns that artificial super intelligences threaten to dominate and disempower us. If people make themselves dependent on AI, they become zombies. Humanity is threatened with “a last narcissistic offence,” says this pop star among the philosophers. Only a system of rules and structures can remedy the situation in order to protect man from manipulation by others and preserve his uniqueness.

On the other side of the aisle is Prof. Dr. Jürgen Schmidhuber who is convinced that artificial intelligence has no interest in enslaving people. The Scientific Director of the Swiss Research Institute for AI IDSIA is regarded as the father of modern artificial intelligence. He welcomes the development with open arms: “I want an AI that learns to solve all problems that I cannot solve myself.”

AI: Dumb machines or dumb people? 

Trend researcher and futurologist, Matthias Horx, denies not only the threat potential, but also the benefits of AI. The head of the Zukunftsinstitut in Frankfurt considers the topic of artificial intelligence to be overestimated. It has become a kind of fetish, with hype that creates dangerous illusions. Most real problems are far too complex to be solved by data systems, explains the sociologist.

Psychology professor, Alison Gopnik, does however, acknowledge the threat – after all, AI is also used to operate weapons. But, “natural stupidity can do far more harm than artificial intelligence,” she writes in the Süddeutsche Newspaper. People need to become more skilled at regulating new technologies.

AI: Ethics or diversity?

Picture shows naked people with helmets and glasses, they look disoriented; smaller part of the picture

Brad Smith’s thoughts also go in the same direction. The President and Chief Legal Officer of Microsoft Corporation emphasizes that it is important for people to decide what computers can do. Therefore, ethical principles should be established. “AI systems must be fair and there must be some kind of accountability for those who develop AIs. Before we adopt new laws to deal with AIs, we need to be aware of the universal values that should be protected by the AI principles,” says the trained attorney in his book, The Future Computed.

Are occupations with a certain pretense also worth protecting according to these principles? In any case, according to Mary Gray, artificial intelligence generates many jobs that are rather tedious. For pattern recognition, AI needs huge amounts of input. The anthropologist, author and researcher at Microsoft Research fears that this input must be recorded by countless so-called click workers. “The greatest paradox of artificial intelligence is that it has a reputation for taking work off our hands. In doing that however, it generates an unlimited amount of new work – repetitive work that isn’t particularly multifaceted,” Gray writes in the Neuen Zürcher Newspaper.

An AI is as helpful or as threatening as the algorithm behind it, stresses Carla Hustedt. The Bertelsmann Stiftung political scientist sees a great danger in a kind of monoculture in algorithmic systems. If, for example, every Human Resources department were to use the same system, the same people would always be discriminated against. “We need a diversity of algorithms,” she writes in an article for the Austrian Standard, “and that also means that it’s not just young white men who design technology.”

More Information: www.t-systems.com/ai-based-automation

Author: Heinz-Jürgen Köhler
Photos: iStockphoto

Do you visit t-systems.com outside of Germany? Visit the local website for more information and offers for your country.