This sponsored post explores the forefront of digital innovation in political discourse. Discover how cutting-edge technology is crafting safer, more engaging online spaces for political dialogue. Interested in featuring a sponsored post on PartyParty or have questions? Contact Marie Wutzler (marie@political.party) for more information.
Nominated as a standout product in the Democracy Technologies category for the 2024 Innovation in Politics Awards, CEO Tomáš Halász talked to Party Party about the genesis of TrollWall AI and its potential to cultivate a secure online environment conducive to meaningful discourse.
In an age inundated with digital content, effective moderation is paramount. TrollWall acknowledges this challenge and confronts it with innovation. By integrating AI technologies, TrollWall has developed a method that not only enhances efficiency but also fosters constructive engagement – a crucial aspect often neglected in politics.
Viewed through TrollWall’s forward-thinking lens, content moderation transcends mere regulation; it becomes a vehicle for nurturing a productive dialogue. Through AI integration, TrollWall empowers social media managers within political institutions to sift through vast volumes of content accurately and swiftly, enabling them to prioritise the creation of meaningful content, while nurturing valuable connections with their supporters.
TrollWall’s pioneering blend of AI and human input heralds a new era in content moderation. It exemplifies a seamless fusion of technological progress and human values, showcasing the potential for transformative change in our societies. Moreover, it shows how AI can enhance human productivity while leveraging technology for the greater good.
“We support that people have freedom of speech, but not freedom of hate”
As a former content moderator for political institutions, Halász observed a significant surge in hateful comments online during Russia’s full-scale invasion of Ukraine. In an increasingly polarised digital landscape, hate has regrettably become commonplace.
“The issue of hate speech is systemic, and social media platforms often struggle to address it effectively”, he asserts.
Beyond the broader geopolitical context, hateful comments serve as a counterproductive distraction from the key message for both supporters and social media managers alike.
Recognizing this trend, Halász sought a more efficient approach to managing his former role. Initially handling moderation manually, he developed TrollWall AI as a means to automate the process. He understood that moderating content on such a large scale is time-consuming and can incur substantial costs, especially during campaigns and elections. Moreover, manual moderation was inefficient, inadvertently normalising hate speech on social media platforms. Moreover, the task of moderation can take a toll on the mental health of moderators themselves, with many experiencing psychological stress and even being diagnosed with PTSD after spending long hours on the job.
It was then that he realised this could be a prime example of a task ripe for takeover by Artificial Intelligence.
“With the help of AI, we can do what we couldn’t do before technically. We can do it now on a large scale”
However, such a transfer of responsibilities raises concerns about bias and the technology’s ability to distinguish genuinely harmful comments. TrollWall employs a unique business model to eliminate political bias from its processes.
TrollWall’s AI is an in-built tool trained with proprietary data, following internal and internationally recognized guidelines on hate speech. The software analyses comments from social networks, written by real individuals, but anonymized to remove any bias around the author’s background. These comments are then annotated by human native speakers, with each comment reviewed by at least three people independently. Furthermore, TrollWall prioritises team diversity to better grasp local contexts and nuances, including regional slang and cultural sensitivities.
After the annotation the data is fed into the AI for training. All the processes are updated weekly while ensuring quality control.
The model ensures TrollWall has full control of the process and removes bias from its decision tree.
“There is always this human overview that the AI is doing the right work”
TrollWall ensures complete transparency with its clients. They have access to all the decisions made by AI and can view all comments that were hidden by AI. Clients can rate and override these decisions at any point and report instances where they believe the technology made an incorrect assessment. So far, Artificial Intelligence has successfully managed over 200 social profiles of various institutions.
“We empower clients to spend time with the voters, not with the haters.”
The tool seamlessly integrates into the PR systems of political parties, enabling quick evaluation of social media comments by AI within seconds of logging in. Comments are then either hidden or retained based on their content. No meta-data of social media accounts is retained during the process, ensuring high data privacy standards.
Clients can undergo a self-onboarding process and receive assistance from TrollWall’s account managers. Specifically tailored for political clients, the tool provides specialised training on utilising the software at scale. Additionally, political clients receive dedicated evaluation and support during the peak of their campaigns.
In addition to operating on platforms like Facebook, Instagram, YouTube, and soon TikTok, the tool boasts a significant advantage in its ability to support live-streamed events, showcasing the power of instant moderation. Leveraging cloud-based software, TrollWall has unparalleled scalability.
The product is currently available in five languages (Slovakian, Czech, Polish, Romanian and German), and Halasz will expand into English next. New vertical features are regularly added based on client feedback. Recent enhancements include advanced analytics of comments, with many more features currently in development.
A tool for saving time and ensuring peace of mind
TrollWall AI has amassed a portfolio of success stories, with its implementation endorsed by political figures such as Slovakia’s President Zuzana Čaputová, Czech Republic’s Petr Pavel, former Slovakian President Andrej Kiska, various political parties across Europe, international NGOs including Amnesty International and Greenpeace, and local authorities.
The software firm takes pride in its track record of never losing a client due to the quality of the product. For certain campaign teams, TrollWall has become the go-to tool for their social media strategy. Furthermore, according to tests, the tool ensures that hiding comments has no impact on the organic reach of social profiles.
Bringing the love back to the campaigns
Halász is gratified to see political parties prioritising online content moderation and engaging in meaningful discussions with voters. He hopes that this shift will lead to more respectful campaigns, transforming them into competitions of ideas and messages rather than hate contests.