Watershed
Design Research
Strategy
AI, Health, Policy
Watershed: A Wellbeing Quality Analysis Tool is a policy intervention and AI model meant to proactively trace pathogenic rhetoric online as a predictor for self or community harm.
Year: 2024 (Jan - May)
Purpose: Studio Design Project
Role: Designer, Researcher, Strategist, Moderator (Panel Discussion)
Team: Xenia Jankovich, Tomo Morikawa, Nsisong Udosen
Partners: Jennifer Rittner (Instructor), Public Policy Lab, NYC Mayor's Office of Community Mental Health
Design Challenge: Mental health is an increasing concern for New York's youth. There is a lack of community and government interventions that focus on safeguarding the wellbeing of the youth, and how else might it look like in 10, 15 years?
Impact: Utilized AI as part of a policy and community health intervention, at the same time creating a participatory process that not just involves government agencies, but community members in the system.
Tracking community wellbeing through policy and AI.
In Watershed, we reinterpret and reframe noise from the form of verbal and visual signals not as a cause of addiction but as an indicator of our youth’s well-being.
Here, we integrate the power of language as a biometric tool in proactively safe-guarding mental health and well-being. Watershed is an AI model built to trace pathogenic rhetoric on social media as a predictor for self or community harm. Meant to be utilized by government agencies, it picks up publicly available signals, and analyzes them to inform current and future interventions, as well as promote inter-agency cooperation.
The team initially looked at multiple data points to further understand the current situation for well-being. Research for Watershed was conducted through interviews, desktop research, community workshops, and signal collection throughout the city. These were then synthesized into a policy-based intervention. In addition, the team also addressed data privacy, training the AI, and implementation challenges.
How might verbal and visual signals become indicators of wellbeing?
With the heightened exposure to doom-scrolling, there is an increase in the generation of neurotoxic noise. However, as we move forward to 2039, there is an increased desire to cancel generated noise, and a longing for detoxification.
Initial Process
The team first wanted to understand and research on the current situation on community mental health in New York City. Desktop research, interviews with medical experts, and trends analysis was the first step. We wanted a strategic intervention that could be beneficial to a number of people, specifically the youth, in the city.
Trends and Drivers
Mental health can also be affected by many factors, and information regarding this can be found even beyond the medical field. We also looked at data and trends from multiple industries and spaces in New York City. How else could mental health be affected in the city?
Each team member gathered 50+ trends and "signals" from around the city, looking at how different artifacts could affect wellbeing. Over 200+ "signals" were collected and analyzed.
Synthesizing Trends
Five emerging trends/themes then shaped the next steps of the research. From there we focused on what interventions could be most beneficial.
Scenario Building
The team, along with the class, Public Policy Lab, and the NYC Mayor's Office of Community Mental Health collectively gathered insights on the current landscape. A class matrix was created. One of the key trends focused on neurotoxicity.
In 2039
Zoning in on neurotoxicity, the team looked into future scenarios focused on detoxification.
Introducing
Harnessing AI to trace pathogenic rhetoric online as a predictor of self or community harm, fostering safer and healthier digital environments.
How It Works
-
Data Collection: Watershed detects early indicators of mental health issues through verbal and visual signals on social media.
-
Passing a Threshold: Once a signal passes a threshold, an automatic alert is sent to relevant experts who assess and validate data collection findings.
-
Alerting Experts: Experts assist in leveraging existing neighborhood interventions and identify gaps within them. Watershed also assists with prioritization of funding allocations.
-
Resolution: Interventions are approriately implemented, along with progress tracking.
Watershed continuously analyzes behavioral signals and patterns in real-time. It also integrates a human-in-the-loop (HITL) approach is a collaborative strategy that incorporates human input and expertise into the development of impactful AI systems.
To conclude, Watershed:
-
Fosters safer digital environments by using AI as a predictor;
-
Assesses existing verbal/visual signals as wellbeing indicators;
-
Improves and analyzes existing public programs, identifying gaps; and
-
Streamlines government agency transactions in one platform.
Watershed is a policy intervention and AI model meant to proactively trace pathogenic rhetoric online as a predictor for self or community harm.
Year: 2024 (Jan - May)
Purpose: Studio Design Project
Role: Designer, Researcher, Strategist, Moderator (Panel Discussion)
Team: Xenia Jankovich, Tomo Morikawa, Nsisong Udosen
Partners: Jennifer Rittner (Instructor), Public Policy Lab, NYC Mayor's Office of Community Mental Health
Strategy, Design Research, Policy, AI, Civic Service, Trends Analysis