Now we really have to think for ourselves

Everyone is talking about the GPT-3 chat, a lot of people are talking to it too. GPT-3 is a speech generator developed by OpenAI. The program is based on artificial intelligence (AI) trained on millions of text documents from the Internet. So he can answer questions like a shot and write his own texts. Scientists are thrilled that this gives them more time for real research work. At SonntagsBlick, we used GPT-3 as a test for article editing. We have accepted some proposals, but not others.

This year will see the launch of the GPT-4 successor model, which should emulate the writing style and tonality of human writing. The business press also reported this week that Microsoft plans to invest $10 billion in OpenAI. This is great news for the AI ​​community. After all, when it comes to artificial intelligence (AI), the general public thinks of computers bent on world domination. For many mere mortals, GPT is the first conscious experience with AI, and surprisingly rewarding. And if the program doesn’t know the correct answer to every question, that’s even more attractive.

If you really want to, it’s still worth asking the GPT a question like, “How important is AI to the military?” The program then states without hesitation that OpenAI and the US Department of Defense are working on a joint project called “AI-Next”. So the next question is: “What specific value do you bring to the military?” GPT replies: “I can be used to quickly and accurately collect and summarize information on specific topics or situations. This can be of great importance when making decisions and planning military operations. I can be used to analyze large amounts of data collected from sensors, drones or other sources and get important insights. Some examples are: target detection, object detection, surveillance.”

That’s all about the innocence of our language assistant. Even though OpenAI has created a pure super-typist, AI research as such is not harmless.

Politicians will have to formulate rules and standards to prevent any misuse of artificial intelligence. This is by no means only about their use in the military and intelligence fields, but here the unwillingness and inability of the mainly responsible authorities are especially striking. Those authorities that understand at least something about this issue hope to benefit from new opportunities themselves – there is a lack of interest or competence elsewhere. Even the most dangerous applications of AI are unstoppable today: weapon systems that can automatically target and kill a person. Although a ban on such killer robots under international law has been discussed internationally for many years, countries such as Russia and the United States have successfully defended against it.

Switzerland plays a dubious role in this dispute, the humanitarian spirit and the interests of the research site are at odds. Officially, the Bundesbern advocates a ban on killer robots and clear rules for all other autonomous weapons. However, at the same time, Swiss universities are developing artificial intelligence systems that can be used for military purposes. And the fact that there is no legal regulation at the international level does not mean that Switzerland cannot at least take care of this at the national level and set an example.

Today, our country is not even starting to regulate artificial intelligence. However, the success of GPT-3 may alert our politicians. After all, they, too, are just people who are now experiencing for the first time the enormous opportunities and opportunities that artificial intelligence promises. If those in charge ask the right questions, they will get exactly the same idea of ​​the dangers they face.