Smart Cities Should be Good Cities: AI, Equity, and Homelessness

Overview 

How can the City of Austin and Good Systems leverage AI to better serve, inform, and empower people experiencing homelessness? People experience homelessness as a continuum, ranging from housing instability to incarceration, couch-surfing to street homelessness, and episodic to chronic. Our goal is to use AI to empower both people experiencing homelessness and the agencies who serve them. We will do this by learning about their specific information and service needs and developing AI-based interventions. Our innovative research will transform wrap-around services for individuals experiencing homelessness in Austin, using AI to ensure Austin is not only a smart city, but also a good city. We will also take into account the COVID-19 pandemic in our research. Public health messaging in this time emphasizes “shelter-in-place,” “stay-at-home,” and “safer at home.” This messaging assumes that everyone has a safe place to shelter, a home. During this crisis, we must consider our most vulnerable populations, including the homeless, not just due to the threat of infection, but also the resulting food insecurity and societal instability. The city’s Emergency Operations Center has created task forces to address various issues including homelessness, lack of income, and food insecurity. Our research can assist with identifying those experiencing homelessness, or at risk of homelessness, and providing services for them during this COVID-19 crisis.  

Initial Findings 

At this stage, we have conducted initial research on the data and collaborative practices of service providers across the Travis County Continuum of Care and the experiences of service users. Initial exploration of this data has also been used to collaboratively formulate hypotheses and direct ongoing work towards AI-driven interventions to aid stakeholders. Service providers must be empowered to address the full continuum of homelessness in a manner that is responsive to shifting needs and circumstances while still benefiting from population-level data and trend analysis.