Meh, this was a survey conducted at 4 different companies. It only proves correlation, not causation. And even then, it only proves it at these 4 companies since they didn't use random sampling from a larger population. Most of the rest of the article is unfounded speculation about the cause of this correlation. Definitely not an impressive study by any means.
The study was performed in Spring 2021 and 2022. The kind of AI used in the study for testing is not specified, but back then it was very likely a poorly built chatbot - nothing like ChatGPT which was released at the end of 2023.
Something seems off here, though I can't quite put my finger on it. Reading though the study, it gives off the vibe of the classic ice cream consumption leads to drowning example from undergrad psych classes.
There is not a word on how the AI system was designed in the study, so no idea how well it worked and if it was frustrating or easy to use. I can totally see someone using chatbots from 2021 and 2022 daily getting depressed :D
