funded by a grant (awarded dec 2025) from the prosocial design network and based on a PDNxRoblox workshop (attended oct 2025), i am working on an a “pause and suggest” intervention that detects potentially harmful or guideline-violating language before a message is sent.
A simulated chat interface testing how real-time AI rewriting affects toxicity and user autonomy in gaming environments.
Launch Prototype →A high-fidelity "War Room" simulation. Tests if professional role obligations (Counter-Attitudinal Role-Playing) override personal political identity.
Launch Prototype →Does labeling a feed as "yours" make you blind to its bias? A simulation testing polarization and the "Algorithmic Self."
Launch Prototype →An interactive experiment testing "Psychological Ownership." Do users accept mismatched recommendations if they "trained" the model themselves?
Launch Prototype →A comparative study between "Hot" (Rigged Game) vs. "Cold" (Recall) induction methods to measure restorative prosocial behavior.
Launch Prototype →A simulation of Automation Bias. It tests how salient heuristics (like a glowing "99% Match" badge) lead users to cognitively offload decision-making, often ignoring contradictory textual evidence. This prototype serves as a modular framework for studying heuristics in digital choice environments.
Launch Prototype → USC Social Networks Adaptation