Tackling AI Control Challenges

Florida Atlantic researchers have developed a new advanced AI framework designed to better handle and access multiple pieces of information to make automated decisions.

Currently, many autonomous systems today operate from a single point of information, instead of in unison with others. Some entities lead, others follow, etc.

This new in-sync approach will make AI more effective and efficient.

“This work fills a crucial gap in the current AI landscape,” said Stella Batalama, Ph.D., dean of the College of Engineering and Computer Science.

Traditional AI models often assume that all decision-makers act simultaneously, with similar influence and information. That simplification doesn’t work well in noisy, resource-limited settings. To address this, Florida Atlantic’s team built a framework in which a “leader” takes action first and “followers” respond. They also added an event-triggered mechanism to reduce how often decisions must be updated, saving computing power and energy while keeping systems stable.

The framework is based on reinforcement learning so that agents can learn from interacting with their environments over time. It was tested in simulation studies that showed it can handle uncertainty, unequal information among agents and limited communication or bandwidth without losing performance. FAU researchers believe this will be especially helpful for infrastructure systems, where delays, variability, and information gaps are common.

“The implications of this research are far-reaching,” Batalama said. “Whether it’s optimizing power consumption across cities or making autonomous systems more reliable, this kind of innovation is foundational to the future of intelligent technology. It represents a step forward not just for AI research, but for the everyday systems we depend on.”

Read the press release.