Marasović develops and rigorously evaluates frontier AI models, such as LLMs, to empower people through effective AI-assisted decision-making, communication, and creativity. “I advocate for simulations where dynamic and complex interactions unfold over time, where risks are evident, and accountability is promoted, as in real-life decision making,” she said. “I will, as a faculty fellow, create a robust testbed to assess the readiness of LLMs and the human-AI collaboration methods developed by my lab, ultimately enhancing both their development and their ability to improve student learning.”