In an exploratory experiment, we test various AI solutions and rely on an iterative editing process under full human supervision.
About 20 students, PhD students and young scientists will come together for a 3-day hackathon and bring data from their current research or dissertation projects. With the help of Large Language Models (LLMs) and various AI prototypes, this data is transferred into initial manuscript versions of scientific research articles. After a multi-stage editing process consisting of iterative human-machine interactions and feedback loops, the manuscripts are then submitted to a renowned scientific journal for review and potential publication.
Tackling global challenges such as climate change or epidemics requires more research and more effective communication of knowledge. However, due to the heavy workload in the university world, this is easier said than done. Can AI provide relief so that researchers can focus on their core work again: research itself? According to a recent study published by Oxford University Press, three quarters of them already use AI routinely, while not even 10% trust the big tech companies when it comes to data privacy and security. So the gap between use and trust is huge.
With this experiment, we as a scientific institution and as a scientific publisher want to jointly investigate the extent to which generative AI helps scientists to save time when writing their research articles while maintaining a high level of quality in a process conducted under human supervision.