Johns Hopkins Selects Five AI Research Projects
The university is funding faculty-led studies to improve AI reasoning and personalization for the 2026-2027 academic year.
Updated on Oct. 22, 2026 in Artificial Intelligence
Live Poll
Do you trust artificial intelligence systems to handle personal information and reasoning accurately?
Johns Hopkins University has selected five faculty-led research projects for its AI Faculty Research Awards. The initiative, funded through the Johns Hopkins University and Amazon Initiative for Interactive AI, supports research and doctoral fellowships.
Why it matters
These projects address critical challenges in AI development, specifically focusing on system reliability and personalized interactions. The research aims to move beyond current limitations where AI models can struggle with inaccurate feedback or flawed reasoning.
The awards cover a one-year term for the 2026-2027 academic cycle. The research includes developing personalized AI assistants that track user knowledge and optimizing AI reasoning to prevent reward hacking.
The players
Johns Hopkins University
A private research university located in Baltimore that frequently leads advanced studies in medicine and technology.
Eugene Yang
A researcher at the university who is building AI assistants capable of tracking and adapting to specific user knowledge.
Luhao Zhang
A researcher at the university focused on enhancing AI reliability through robust optimization techniques.
The details
Eugene Yang is working on personalized assistants that maintain transparent knowledge records to allow user corrections, while Luhao Zhang is studying robust optimization to stabilize AI reasoning. These efforts target the persistent problem of reward hacking, where systems prioritize high scores over accurate conclusions during training.
Timeline
October 8, 2026: Researchers presented their work at the AI2AI Research Festival.
2026-2027: This period marks the duration of the faculty research awards initiative.
The Tech Race
This effort follows the collaborative framework established by the Johns Hopkins University and Amazon Initiative for Interactive AI to support advanced research. It positions the university at the forefront of the race to create more reliable, transparent, and user-centered AI models.
These developments could lead to more accurate AI tools that better understand individual user needs while avoiding the errors common in current models. Users may eventually experience more reliable and personalized digital assistants in their professional and academic workflows.
The takeaway
Researchers are tackling the fundamental issue of reward hacking to ensure future AI systems provide more accurate outputs. Readers can expect AI tools to become increasingly customized as these optimization strategies move from the lab into practical software applications.
Further reading
Learn more about the latest developments in the field at Artificial Intelligence.
Source note: This article includes information reported by The Johns Hopkins News-Letter.
Live Poll
Do you trust artificial intelligence systems to handle personal information and reasoning accurately?
