The reproduction of social inequalities through artificial intelligence and large language models (LLMs) has been empirically demonstrated in various areas of society. However, there is a lack of a broader discussion on the extent to which LLMs can influence the scientific area, reinforce or mitigate existing structural inequalities and introduce a "bot mania" in science. The authors focus on publication and citation behavior and develop a thought experiment on the effects of LLMs. This distinguishes between the reproduction of pre-existing structural inequalities in academia (socio-cognitive stasis), or a catharsis that can counteract structural inequalities and Matthew effects. Three scenarios for the consequences of using LLMs for citations are presented and developed in this text.

link to text