What is all this about AI?

Why has artificial intelligence become such a hot area right now?

Jan Ljungberg, professor of informatics, looks back in time and reflects on what conditions that have led to the development that we have today. What is it that sets our time apart from previous AI hypes?

– When I was a student in the master's programme in computer linguistics thirty years ago, AI was hot. In 1984 came the book The Fifth Generation, which warned of Japan's big investment in a research programme connected to AI: "The fifth generation will shift every cornerstone of our societies". At this time, all large companies in Sweden such as Volvo and Ericsson AB, had an AI department of its own. The neural networks already existed, but most of them were what we call GOFAI, Good Old Fashioned AI, in terms of expert systems. But for various reasons, the neural networks didn’t get a breakthrough. In 1991 I was at the big AI conference IJCAI, where parts of the Japanese fifth generation project were be presented. Of the original big plans there had not become much. A few years later, the AI hype was more or less stone dead and remained so for at least a decade.

– Today we once again see a huge wave of AI interest, reaching all the way up to government level. Primarily, it is the further development of the neural networks that has achieved a breakthrough in different types of self-learning systems. You hear the same predictions as in the late 80's: AI will fundamentally change our lives. In what way it will do that, there are many opinions and in this short text I won’t go further into them. But it is important to try to understand and predict the consequences of AI development in order to make conscious design choices.

– Will AI deliver this time, or will the AI hype soon fade away? What are the differences now compared to the last hype? There are undoubtedly many similarities and also some naïve commitments.

However, the conditions are different in many ways today:

• The power of the processors: today there is more data power in a small cell phone than in the most advanced AI computers of the time for the first AI hype.
• Access to data: The AI systems using machine learning need to have access to very large amounts of data. Today, digital platforms, sensors and the Internet of Things generate data in quantities you couldn't even imagine thirty years ago. It is also these platforms that many people see as the future giants within AI: Google, Facebook, Amazon and Microsoft, but also the Chinese platforms Baidu, Tencent and Alibaba. In the book AI Super Powers, China is highlighted as the future superpower of AI.
• Digitalization: The comprehensive digitalization of the entire society that has taken place during the last decade has made us more susceptible to new technology.
• The use: Rather than large technological breakthroughs, the use of AI in new problem areas creates the power of the current AI wave. We all use everyday products that include some kind of AI, without thinking so much about it.

– A few of the enormous opportunities and challenges created by the AI development are expressed in the other articles and interviews found on this web page. While we take advantage of the AI potential to solve small and large problems, it is important to try to understand and predict the consequences of AI development in order to make conscious design choices. Researchers from many disciplines – as well as developers, entrepreneurs, policymakers and users – all need to contribute to this development.

Text: Jan Ljungberg, professor of informatics, Department of Applied Information Technology