Since its launch in November 2022, ChatGPT has rapidly gained users and challenged traditional information tools such as Google and Wikipedia. While AI-powered tools like ChatGPT offer fast and accessible answers, they are not infallible, which raises concerns about misinformation. This leads to the question of whether people actually trust ChatGPT compared to its competitors. And additionally: Does a warning about its potential inaccuracies affect this trust?

 

Our research explored these questions through a large-scale survey experiment, analyzing trust in ChatGPT, Google, and Wikipedia. We examined six trust dimensions: three traditional ones (ability, benevolence, and integrity) and three adapted for technology (functionality, helpfulness, and reliability).

 

The findings reveal significant differences. Google was perceived as more functional and reliable than ChatGPT, but both were seen as equally helpful. ChatGPT, however, was considered more helpful than Wikipedia, with no major differences in functionality or reliability. Interestingly, the classic trust dimensions did not show significant variation between the three information tools.

 

Beyond these comparisons, we found that frequent and long-term use of a tool positively influences trust. Additionally, individuals more inclined to trust new technology generally have higher confidence in ChatGPT. Surprisingly, warning the users that ChatGPT itself states on its platform that it may provide incorrect information had no measurable effect on trust.

 

As AI-driven tools continue to evolve, understanding how users trust and engage with them is crucial. While ChatGPT has not yet surpassed Google in perceived trust, its growing role in information gathering suggests that the landscape of digital trust is shifting.