Leonora Tima, inspired by a family tragedy involving gender-based violence, has developed the South African mobile app Grit. The app, which features an AI chatbot named Zuzi, aims to provide support and tools for individuals experiencing abuse, including evidence gathering. Grit's development involved input from communities around Cape Town and is financially and technically supported by organizations such as Mozilla and the Gates Foundation.
Development and Inspiration
Leonora Tima initiated the development of Grit following the killing of her 19-year-old, nine-months pregnant relative in 2020 near Cape Town. The perpetrator was not identified. Tima noted that her relative's death was not reported by news outlets, attributing this to the high volume of similar cases in South Africa. Her observations led to the creation of Gender Rights in Tech (Grit), featuring the Zuzi chatbot. The app is described as an African solution, co-designed with African communities, and is intended to offer support and facilitate evidence collection for potential legal actions.
Grit's Features and Reach
Grit is a free mobile app, though it requires mobile data for download. The team reports 13,000 users and approximately 10,000 requests for help in September. The app is built around three core features:
- Help Button: A circular button on the home screen that, when pressed, records 20 seconds of audio and alerts a private rapid-response call center. A trained operator contacts the user, and if immediate help is required, a response team or a local organization is dispatched. Minor instances of users testing this feature have been reported, but not system abuse.
- The Vault: A secure, dated, and encrypted digital space for storing evidence of abuse, such as photos, screenshots, and voice recordings. This feature aims to protect crucial evidence from loss or tampering.
- Zuzi Chatbot: Launched this month, Zuzi is an AI-powered chatbot designed to provide guidance and direct users to local community support. User feedback indicated a preference for Zuzi to embody an "aunt figure" – warm, trustworthy, and non-judgmental.
During testing, Zuzi has been utilized by men seeking help, including perpetrators requesting assistance with anger management and male victims of violence. Leonora Tima observes that users may prefer interacting with AI due to a perceived absence of judgment.
Gender-Based Violence in South Africa
UN Women reports that South Africa experiences some of the world's highest levels of gender-based violence, with a femicide rate five times the global average. Between 2015 and 2020, South African police data indicates an average of seven women were killed daily.
Expert Perspectives and International Attention
Lisa Vetten, a specialist in gender-based violence in South Africa, acknowledges the inevitable role of technology in addressing GBV but advises caution regarding AI in trauma-centered care. She refers to AI chatbots as Large Language Models and notes instances where they have provided incorrect advice, particularly concerning legal issues. Vetten stresses that chatbots are not a substitute for human counseling, emphasizing the need for trust and safety with other human beings.
Grit's approach has garnered international interest. In October, Leonora Tima's team presented the app at the Feminist Foreign Policy Conference in Paris, where global leaders discussed technology's role in achieving gender equality. At the conference, 31 countries pledged to prioritize tackling gender-based violence.
Lyric Thompson, founder of the Feminist Foreign Policy Collaborative, notes challenges in incorporating gender-related biases into AI discussions. Heather Hurlburt, an associate fellow at Chatham House specializing in AI, states that AI has the potential to either address gender discrimination and violence or entrench misogyny, depending on its development.
Leonora Tima asserts that the success of AI in combating gender-based violence depends on who designs the technology. A 2018 World Economic Forum report indicated that only 22% of global AI professionals were women. Tima highlights that current AI is built with historical data primarily centering male voices, particularly white men. She advocates for broader representation among creators, including women of color, individuals from the global south, and those from less privileged socioeconomic backgrounds, to ensure technology reflects user realities.