Using CI to mitigate against the negative impacts of AI on online communities
Almost all online social networks make use of AI within their platforms, yet the full extent of the algorithms’ impact on the communities that use the platforms are still rarely rigorously evaluated in real-world settings. There is evidence to show that negative behaviours of users on popular social networks like Reddit and YouTube are amplified by the use of AI, which can automatically promote the most attention-grabbing or shocking content. Experiments into these trends tend not to be co-designed with members of online communities, even though these individuals hold unique insights about how the platform is used and are important for driving behavioural change that can improve users’ experience of online environments.
The mission of the Citizens and Technology Lab (CAT Lab) at Cornell University is to ‘work with communities to study the effects of technology on society and test ideas for changing digital spaces to better serve the public interest’. By involving existing online communities in the research, the group can ensure that they pose research questions that are the most relevant to the real-world setting.
Using a collaborative citizen science model, the academics work together with an online crowd to set research questions and design experiment protocols. In 2016, CAT Lab ran a study to test the effect of different community moderation practices on mitigating against harassment and combatting the spread of fake news, in forums hosted by the Reddit platform.
The research showed that online communities can help to consolidate positive social norms through simple actions, such as regular reposting of community rules in discussions. These interventions increased rule compliance by 8 per cent and significantly increased contributions by new users, both of which can serve as a counterpoint to the negative content promoted by AI. Designing and testing interventions with the communities most affected helps to instil more positive behaviour on social platforms and empowers communities to take part in the governance and monitoring of new technology.
The CAT Lab uses CI to evaluate the impact of digital technologies, including AI, on social interactions online. Since its launch in 2014, the group has completed five studies on some of the most popular online platforms, including Facebook, Reddit and Twitter.
 Formerly known as CivilServant.