(MENAFN) A groundbreaking artificial intelligence system developed by researchers at Duke University can now tackle sophisticated design challenges with near-human expertise, potentially revolutionizing scientific discovery across numerous disciplines, according to findings released last week in ACS Photonics.

The breakthrough involves an "agentic system" powered by large language models that function collaboratively as digital research teams. These AI agents operate autonomously to engineer and evaluate metamaterials—artificially constructed substances with specialized electromagnetic characteristics—requiring zero human intervention throughout the process.

The intelligent network distributes distinct responsibilities among individual AI units: data management, neural network programming, accuracy validation, and performance enhancement. A master algorithm orchestrates inter-agent collaboration while tracking overall advancement.

Performance evaluations revealed the system's best outputs rivaled those produced by seasoned human researchers, although its median results fell marginally short of expert-level work.

Scientists behind the innovation assert the findings demonstrate that sophisticated AI architectures, when properly configured, can autonomously navigate exceptionally intricate scientific puzzles. They project these agent-based AI platforms will rapidly augment—or potentially replace—traditional research methodologies in disciplines spanning materials engineering to chemical sciences.

The US-based research team's work signals a pivotal shift toward AI-driven scientific experimentation, where machine learning systems could soon independently conduct complex investigations previously reserved for trained specialists.

MENAFN26102025000045017169ID1110248225