The question of whether it is possible to simulate thinking with machines (Turing 1950) or whether it is possible for machines to think (Minsky 1961) has been debated for several decades (Weizenbaum 1976; Searle 1980). While on the one hand, due to the identity of assembly instructions and opcodes, it is clear that the possibility of thinking machines does exist neither theoretically nor practically, the belief that computers can potentially think better than humans and that this has already been achieved in some fields is maintained. This is especially true for adherents of the data religion (Harari 2017).

The question of possible uses of systems based on artificial intelligence methods in scientific culture is also discussed (Krenn et al. 2022). In this debate, the assumption is argued that AI systems will write scientific papers in the future (Gil 2021, 13). However, no examples of AI systems acting as "agents of understanding" (Krenn et al. 2022, 767) have been reported so far. One such example was examined for this talk.

The selection of the example was motivated on the one hand by the fact that submitting papers for conferences and reviewing papers for conferences is currently a widespread part of scientific culture. At the same time - not least under the influence of the Californian ideology (Barbrook and Cameron 1996) spread by adherents of the data religion - more and more lectures, publications and external funding are demanded of scientists. This necessitates an increase in productivity, and it is an obvious option to test whether current AI systems can contribute to an increase in productivity in the area of talks, in other words, whether the problem can be solved by the means that cause it.

The rather low-threshold genre of the poster (D'Angelo 2016) was chosen for the experiment. Two poster proposals for the ECER meeting were generated and submitted using ChatGPT. In addition, peer review of papers submitted for ECER was conducted using ChatGPT. These reviews were not submitted for factual and ethical reasons, but were compared with reviews generated by humans for the same contributions.

The results of the reviews of the lecture proposals made with ChatGPT and the comparison of the reviews made by ChatGPT with the reviews made by humans will be presented. It is expected that the study will show that the lecture proposals are rejected, that the appraisals fail to meet the proposals and that AI systems cannot currently achieve any work facilitation in the pedagogical scientific culture in this respect.

D’Angelo, Larissa. 2016. Academic posters: a textual and visual metadiscourse analysis. Linguistic Insights, volume 214. Bern ; New York: Peter Lang. Gil, Yolanda. 2021. „Will AI Write Scientific Papers in the Future?“ AI Magazine, Nr. 42: 1–15. Harari, Yuval Noah. 2017. Homo Deus. Eine Geschichte von Morgen. München: C. H. Beck. Krenn, Mario, Robert Pollice, Si Yue Guo, Matteo Aldeghi, Alba Cervera-Lierta, Pascal Friederich, Gabriel dosPassosGomes, u.a. 2022. „On Scientific Understanding with Artificial Intelligence“. Nature Reviews Physics 4 (12): 761–69. https://doi.org/10.1038/s42254-022-00518-3. Minsky, Marvin. 1961. „Steps toward Artificial Intelligence“. Proceedings of the IRE 49 (1): 8–30. https://doi.org/10.1109/JRPROC.1961.287775. Searle, John R. 1980. „Minds, Brains and Programs“. Behaviroal and Brain Sciences 3 (3): 417–57. Turing, A. M. 1950. „Computing Machinery and Intelligence“. Mind LIX (236): 433–60. https://doi.org/10.1093/mind/LIX.236.433. Weizenbaum, Joseph. 1976. Computer Power and Human Reason: From Judgement to Calculation. First Edition. San Francisco: W.H.Freeman & Co Ltd.