AI: The dark side of power?

When we think of the dangers of artificial intelligence, the first thing that comes to mind is the on-board computer HAL from the movie "2001 A Space Odyssey". But even without the omnipotence fantasies of science fiction machines, artificial intelligence already has a dark side that is still completely unknown to many.

Investors love artificial intelligence. Start-ups that use AI technologies therefore get money more easily. This is probably the reason why 40 percent of all AI start-ups in Europe use the term AI without actually using AI technologies. We owe many of the advances under the buzzword "machine learning" not only to technological innovations, but also to new opportunities that allow us to exploit human labor on digital platforms worldwide. Artificial intelligence requires armies of workers who use human intelligence to process, evaluate and compile data at great expense. This data is then the input for the machine learning algorithms.

Some AI companies use the "fake it till you make it" approach and rely on human labor because the performance of the AI algorithms is often not sufficient for the full-bodied marketing announcements. Behind many a hip AI app is, at the very end, a human being. These people, often without an employment contract and earning cents per click, remain invisible to us; they are the so-called ghost workers and form the new AI precariat. They exist worldwide, in industrialized countries and especially in developing countries. Without them, the global AI industry would no longer function. The training runs with data consume huge amounts of computer power - with corresponding energy consumption and CO2 emissions. The cost of a single training run for large AI models can quickly run into the millions of euros. Smaller companies and start-ups can no longer keep up.

Deep fakes, cybercrime and digital deception

When Barack Obama called Donald Trump a "dumbass" in a video, he wasn't really doing that, of course. The deceptively genuine video is a deep fake, a video manipulated with the help of AI. There is a risk that it will soon no longer be possible to distinguish between manipulated and genuine material. Such a development could have a massive impact on the media, companies, law enforcement and ultimately even democracy. A study by Europol found that cyber criminals are now exploiting victims more easily and quickly through the use of artificial intelligence. There are algorithms that guess passwords and falsify photos, videos, sounds and voices as desired. AI also makes known hacks more powerful and at the same time helps to better conceal the attacks.

Surveillance, discrimination and the need for regulation

Whether selection through facial recognition, automatic exclusion from platforms, political repression or discrimination: AI makes it easier for people to put themselves above others. According to an article in the New York Times, China is using a facial recognition system to technologically monitor the Muslim minority of 11 million Uyghurs. This makes China the first country in the world to use artificial intelligence not only for economic but also for political purposes.

The advantages of artificial intelligence are obvious. Little is known to the public about the dark side of AI, the shadowy side that exists every day behind the otherwise shiny façade of AI. AI is already being used in manipulative, discriminatory, surveillance and criminal ways. I therefore consider the current European efforts to regulate AI to be absolutely sensible and necessary.