Humour is unattainable for artificial intelligence

Artificial intelligence (AI) cannot understand humour and is unable to reproduce jokes and memes, reports Week magazine.

A study based on a review of scientific papers says artificial intelligence is unlikely to be able to understand humour and learn how to create jokes. AI researchers around the world largely agree that AI can find itself in ridiculous situations and recreate laughter completely by accident, but a machine can't replicate comedy production on a conveyor belt. According to University of Delaware scientist Ishaani Priyadarshini, in order to understand even the simplest internet meme, AI must go through several levels of understanding. The specialist is convinced that jokes and memes, due to their subjectivity, are unattainable for AI, and in this, humans are vastly superior to machines. Furthermore, comedy routines often transcend the boundaries of decency and political correctness, whereas AI is a conservative technology that adheres to strict rules.

In theory, robots and AI models can learn to write jokes, according to University College Dublin professor Tony Villa. To do so, they will have to work with absurdity, which is considered the basis of humour. Such a model would be programmed to identify inconsistencies and deviations from established standards.

At the end of June, scientists at the University of Oslo (Sweden) found that the human brain can distinguish real voices from synthesized ones. However, it is almost impossible to determine by ear the voice of an AI. | BGNES