By Gabriella Gricius, Jasmijn de Zeeuw, and Bethany Houghton - PILPG-NL Senior Research Associates
Over the past six months, PILPG’s Netherlands Office has been working with Vrije Universiteit Amsterdam, the Knowledge Management Fund, InterCidadania Institute, the Legal Assistance Centre in Namibia and the VU Vereniging on the Virtual Human Rights Lawyer (VHRL) Chatbot. One of the main reflections from this project as well as the takeaways from our expert input sessions was how many challenges we first faced and later have overcome.
At the beginning of this process, our team put together a list of some of the challenges that building the VHRL chatbot presented. For some of these challenges, we devised ways of meeting them, while for others we have adapted our approach to better take these factors into account. After much reflection at the end of the project period, we have determined some ways of adapting and meeting the most critical of these challenges.
Challenge 1, 5 & 6: Internet, Adaptation to Local Customs, and Interacting with Victims
The first and most obvious challenge of the VHRL chatbot was that it relies on internet access. Victims of human rights violations often do not have internet access and therefore, a lack of internet access was a severe limitation of the chatbot’s usefulness. Furthermore, adapting the chatbot to local language presented another issue as many different cultures talk about certain issues such as rape in a roundabout way. Lastly, the challenge of how to interact with victims was one that we spoke to numerous experts about.
To solve these three issues simultaneously, our team decided that to refine the focus on creating a tool for legally-minded civil society organizations (CSOs) rather than focusing on directly creating a tool for victims. Not only does working with legally-minded CSOs partially solve the challenge of creating a chatbot with understandable but legally accurate language, it also means that it is more likely that our target audience would have internet access. This ensures enhanced accessibility of our tool.
Challenge 2 & 7: Building the Chatbot and Data Protection
Two separate but interrelated challenges are the actual construction of the chatbot as well as the question of data protection. Constructing the chatbot was a challenge that our team dealt with by working with InterCidadania Institute to create a first prototype. This version of the chatbot brought many further challenges to the forefront such as reflecting on whether we should be working with ontological or decision tree designs and how to create a sustainable backend that could be modified to use natural language processing once this technology is ready.
The second challenge for our team was data protection. How would we make sure any data was securely stored? Did we want to collect data at all? Would the data that ran through Telegram as a third party be safe? Victims of human rights violations are particularly vulnerable and their data highly sensitive. Protecting this data is central to the development of the chatbot. To approach this challenge, we looked closely at data protection regulations throughout Europe and varying ways we could ensure that data would remain safe. We also examined at the possibility of not collecting data at all. Looking forward, we remain committed to making data protection and cyber security the most important aspect of the development of the VHRL chatbot.
Challenge 4: Domestic Remedies
The VHRL chatbot focuses on international and regional human rights mechanisms. These mechanisms in manyl cases ask applicants to first use all judicial proceedings available under domestic law before bringing a case or complaint before them. This poses a challenge for our chatbot - how can we ask users if they have exhausted domestic remedies? Which questions can help determine whether this is the case when domestic systems vary? And how to incorporate the exceptions to this requirement?
Our choice for legally-minded CSOs addresses this challenge. The VHRL can assist those CSOs that work with clients who have already exhausted domestic remedies and are looking for another way to help them with their case. The chatbot aims to make information about mechanisms at the international level - information that is currently dispersed and often difficult to access - more accessible for these (local) organizations.