Summary
- OpenAI and MIT conducted a 28-day longitudinal study on CloudResearch Connect
- 981 participants completed > 32,000 surveys in a randomized controlled trial
- Each day required actual interactions with ChatGPT and a post-interaction survey
- Participants completed pre and post survey measures of emotional well-being, plus a survey at the end of each week
Over 400 million people use ChatGPT every week. When OpenAI, wanted to know how regularly using ChatGPT affects people’s emotional well-being, they chose CloudResearch Connect. This blog explains why.
Why OpenAI Runs Studies on CloudResearch Connect
Early in 2025, researchers at OpenAI and MIT wanted to understand how regularly using ChatGPT affects people’s emotional well-being. To do so, they designed a randomized controlled trial that spanned 28 days. The study required about 1,000 participants to complete 28 daily touchpoints, 4 weekly surveys, and 2 comprehensive assessments—one before and one after the trial period. That is 34 interactions per participant.
Perhaps it goes without saying, but this is a lot to ask of participants. It’s also the kind of research that faces numerous practical challenges.
To conduct a study with more than 32,000 data collection points, the researchers needed to find participants who were willing to complete daily tasks for multiple weeks, not people who sign up and then disappear. The researchers also needed a way to directly communicate with participants, to remind them of their daily sessions and weekly surveys. Perhaps most of all, the researchers needed confidence that the data would be trustworthy and able to answer the research question.
OpenAI chose CloudResearch Connect because the platform offers all of the things needed for a study of this complexity: Longitudinal projects with retention rates of 80% or more; vetted participants with a proven track record of engagement; the technological infrastructure required to communicate with participants and track their progress; and a reputation for high quality data that helps answer pressing research questions.
How Did the Study Work?
The OpenAI study was structured as a randomized controlled trial with multiple interaction modes. Participants were assigned to either voice-to-text or text-only interactions with ChatGPT, and some were given specific topics to explore.
Over 28 days, the participant journey looked like this:
- Day 0: Baseline assessment
- Days 1-28: Daily ChatGPT sessions (5 min minimum) + post-session evaluation
- Days 7, 14, 21, 28: Weekly well-being survey
- Day 28: Post-study assessment
- Total Time Commitment: > 200 minutes per participant
- Total Compensation: $100 per participant
Managing this level of daily engagement across 1,000 participants represents a significant logistical challenge. But several features of Connect contributed to the study’s success.
What Made This Study Successful?
The success of OpenAI’s study on Connect came down to several key factors that align with best practices for longitudinal online research.
Vetted, High-Quality Participants
CloudResearch Connect maintains its reputation for data quality through careful participant vetting. Rather than accepting anyone who signs up, Connect screens participants to ensure they provide reliable, thoughtful responses. For a 28-day study requiring daily engagement, this vetting process was crucial. OpenAI could trust that participants who enrolled would actually show up day after day and provide quality responses.
Participant Targeting and Selection
The study required targeting specific types of participants, including considerations about prior ChatGPT usage. CloudResearch Connect’s participant database and targeting capabilities allowed OpenAI to recruit the right sample for their research questions.
Built-In Communication Systems
One of the challenges of any longitudinal study is keeping participants engaged over time. CloudResearch Connect’s infrastructure allowed OpenAI to send daily reminders and maintain consistent communication with participants throughout the 28-day period. This kind of sustained engagement is difficult to achieve without the ability to directly communicate with participants.
Fair Compensation
OpenAI compensated participants appropriately for their time. Each participant who completed the full study earned $100. The resources OpenAI invested into this study not only ensured people were motivated to complete the study, but they also communicated that the researchers valued participants’ contributions and their effort. They likely improved data quality, too.
The Future of AI Interaction Research
As artificial intelligence becomes increasingly integrated into daily life, understanding how people interact with AI tools over extended periods becomes more critical. OpenAI’s study on CloudResearch Connect demonstrates that ambitious, longitudinal research on human-AI interaction is not only possible but can be conducted with rigor and reliability.
If you’re interested in conducting longitudinal research online, CloudResearch can help. Our Connect platform combines vetted, reliable participants with the tools needed to maintain engagement and ensure data quality. In most longitudinal projects on our site, researchers achieve retention rates of 80% or more! Contact us to learn more about how we can support your research.
Did You Know?
The OpenAI study adds to CloudResearch’s portfolio of successful longitudinal projects, including the most intense longitudinal study ever conducted online, which we did in collaboration with IARPA. All of the lessons we’ve learned and the best practices for conducting online studies are built right into our Waves system, which reduces the time required to set up and manage longitudinal studies online.
