AI can help humanitarian actors gain critical insights to better monitor and predict risks such as the emergence or escalation of conflict. But deploying the systems in this context is not without risks to affected people, new research warns.
Humanitarian organizations are increasingly using digital technologies, and the COVID-19 pandemic has accelerated this trend.
AI-powered disaster mapping has been used to speed up emergency response in Mozambique, and AI systems have been used to predict food crises and deployed by the World Bank to 21 countries.
However, the study warns that some uses of AI could cause further harm to people and pose significant risks to the protection of people's rights.
The research, by Professor Anna Beduski from the University of Exeter's School of Law, appears in the Handbook of War and Artificial Intelligence.
“AI technologies have the potential to further expand the humanitarian preparation, response and recovery toolkit,” Professor Beduski said.
“But safeguards need to be put in place to ensure that AI systems used to support humanitarian efforts are not transformed into tools to exclude people in need. Safeguards also need to be put in place to respect and protect data privacy.”
“Any deployment of AI systems in situations of conflict or crisis must be guided by the humanitarian imperative to 'do no harm.'”
The study finds that humanitarian organisations designing AI systems should ensure data protection by design and by default to minimise the risk of harm, regardless of whether they have a legal obligation to do so, and should use Data Protection Impact Assessments (DPIAs) to understand the potential adverse effects of these technologies.
Grievance mechanisms should also be established to allow people to challenge automated decisions, or decisions made by humans with the assistance of AI systems, when they adversely affect them.
Professor Beduski said: “AI systems can analyse large volumes of multi-dimensional data at increasingly fast speeds, identify patterns within the data and predict future behaviour. This can provide organisations with key insights to better monitor and predict risks such as the emergence or escalation of conflict.”
“However, deploying AI systems in humanitarian settings is not without risks for affected people. Issues include the poor quality of the data used to train AI algorithms, the presence of algorithmic bias, a lack of transparency in AI decision-making, and widespread concerns about respecting and protecting data privacy.”
/Public Release. This material from the originating organization/author may be out of date and has been edited for clarity, style and length. Mirage.News does not take any organizational stance or position and all views, positions and conclusions expressed here are solely those of the authors. Read the full article here.
