BERLIN taz | Research on Artificial Intelligence (AI) has reached a peak and is currently booming like no other field of science. The number of scientific AI publications has increased on average by around 12 percent annually in the past five years (2013 to 2017), according to a study by the science publisher Elsevier . In the previous five years, growth was just under 5 percent.
In 2017, more than 60,000 AI research papers were published worldwide, compared to just under 10,000 in 1998. The research location China has an increasingly large share in them. It is now 24 percent, ahead of the USA (17 percent) and behind Europe, which is the strongest AI research region with 30 percent, but decreasing: 20 years ago, it was 35 percent. In the country ranking, China is in third place ahead of the USA and rising star India. Germany comes sixth behind Japan.
600,000 documents and more than 700 AI-specific keywords from the four areas of research, education, technology, and media were evaluated for the Elsevier study. The combination of “extensive text mining and semantic literature analysis” is “currently the most comprehensive analysis of existing AI research,” said Enrico Motta, professor of knowledge technologies at the Open University in Great Britain and co-author of the report. “We basically used AI to define AI.”
Most publications were in the research areas of machine learning and probabilistic thinking, neural networks, and computer vision. Other AI fields are search technologies, language processing, and knowledge representation.
The data evaluation also shows interesting hikes. For example, western AI research has to struggle with brain drain to a large extent, the Elsevier study notes. The study says that private sectors in USA appeal to most foreign and local students of the sciences to AI talent and learning while the European private sector is dependent upon the non-European private sector. The reason for this is the better pay that companies can offer researchers hired from universities.
The relationship between artificial intelligence and ethics was also queried. According to the Elsevier report, it was shown that “ethical issues are much more important in society and the media than in science”. It is not out of the question that there will be an acceptance problem for AI researchers.