Philipp Marten @plmarten.bsky.social · 11/07/2025ChatGPT outperformed human participants in source discernment but only when all relevant information was clearly available. 💡 It struggled when sources were less credible or information was not available in the materials. 110 1
Philipp Marten @plmarten.bsky.social · 11/07/2025So, what does this mean? 🤖 AI tools like ChatGPT may assist with sourcing but they still fall short in critical information evaluation. They cannot - and should not - replace human judgment in online reasoning. 110 1
Philipp Marten @plmarten.bsky.social · 11/07/2025Special thanks to my supervisor @marcstadtler.bsky.social and Mykola Makhortykh for their collaboration as well as Deutsche Telekom Stiftung for supporting this research. 110 1
