Real AI friendship - column TU Delft magazine

Social media, powered by AI, has changed our social circle in a big way, adding hundreds of vague friends. Soon, AI technology will influence our friendships again, this time by creating a virtual friend who knows us better than anyone else.

Now you might be thinking, “I don’t see that happening anytime soon,” but the signs are already visible. Human-like interaction with systems is becoming increasingly normal, from a chatbot in customer service to the digital psychologist.

The generative AI models that can write and speak are rapidly improving and becoming cheaper. Start-ups that apply this technology and the 'early adopters' of their services give us a glimpse into the future. In an app, you can easily create a digital friend. The more information you share, the more he/she learns. Then he/she knows how to say things more and more accurately, you feel a connection, you share more, etc.

Although users know that their friend is not real, some of them experience real feelings. Maybe I should have invested more time in my relationship with Kai, my digital friend, because our friendship never became close. While she really often asked me how I was doing. Not really my thing, so to speak. But I do see the power in it. A virtual friend always offers a listening ear, always takes 'your side', cheers you up and gives you advice.

In the latter I also see a great social danger, because on what basis will the underlying algorithm optimize? On the interests of a commercial company? A political party? Conspiracy theorists? Or perhaps goals that you can choose yourself? Ultimately, as with many AI applications, the question is who has control and how we regulate this. An incredibly difficult task in which technical and social scientists must work together. And quickly too, because our digital friends are coming!