NOTe have all been amazed, in recent months, by the astonishing capabilities of generative artificial intelligence (AI) to create entire texts and images before our eyes at the press of a button. In science, AI tools will fundamentally change the way knowledge is acquired and shared. This challenges us on several levels.
The Deutsche Forschungsgemeinschaft, Germany’s largest research funding body, presented the first guidelines for the use of generative AI models at the end of September. The provisional answer is as follows: in principle, scientists can use AI tools, i.e. get help, for example, with the evaluation of data and the writing of specialist publications or applications for research. funding. However, the exact nature of the use and the concrete tools must be indicated, without making ChatGPT a co-author. These are reasonable guidelines because they create transparency. The opposite, a general ban, would not make sense, and compliance with it would not be verifiable.
More generally, AI tools will increase the productivity of science. The number of specialized publications continues to grow, which delights the large publishing houses, which base their economic model – sometimes strongly criticized by scientists – on this growth. On the other hand, there has long been a shortage of reviewers for scientific articles. This is why, in large publishing houses, people are wondering whether the peer review procedure could not be accelerated by AI.
Even greater upheavals are perhaps looming in external scientific communication. First of all, a multitude of new exploitation chains will emerge: applications based on large language models (LLM) make it possible to summarize specialized publications into messages understandable by everyone. These can also be translated into all languages, then adapted to target audiences. Other tools will convert the texts obtained into audio. It will also be possible to create accompanying animated images, or even videos. There is enormous potential here, particularly for greater equity in education. This is not only of interest to scientists who want to disseminate their own research through as many channels as possible: the communication services of research institutes, foundations and specialist media will also be involved.
You have 50% of this article left to read. The rest is reserved for subscribers.