A new review in Medicine Plus traces a century of deep brain stimulation (DBS), explaining how the therapy alters pathological activity from local circuits to whole-brain networks, and mapping targets and clinical evidence across 16 neurological and psychiatric disorders. It also examines adaptive stimulation, directional leads, and biomarker-guided approaches that could make treatment more precise and personalized.

-  Professor Min-Soo Kim’s research team in the School of Computing develops SafeQL, a technology that identifies and selectively corrects errors in AI-generated SQL queries

-  SafeQL resolves execution errors in up to 87.4% of initially erroneous queries in international benchmark tests, while reducing token use by a factor of up to 15.1 and correction time by a factor of up to 29.6 compared with full query regeneration

-  The technology corrects references to nonexistent database tables, columns, functions, and values, helping reduce enterprise AI operating costs and advance reliable AI work assistants

Seventy years after a small group of scientists gathered at Dartmouth College and helped launch artificial intelligence as an academic discipline, AI has evolved from an ambitious idea into a powerful force reshaping science, medicine, industry, and environmental protection.

Three faculty members at University of Tennessee, Knoxville, are among the recipients of 2026 R&D 100 Awards. Jack Dongarra, professor emeritus and Distinguished Professor of Computer Science, and Yilu Liu, UT-Oak Ridge National Laboratory Governor’s Chair for Power Grids, are being honored for their work on innovations that will benefit large-scale scientific computing and protect critical infrastructure. Parans Paranthaman, who holds a joint appointment at Oak Ridge National Laboratory and UT, has been recognized as the 2026 R&D 100 Researcher of the Year. 

The complex properties of spin glasses inspired aspects of modern AI software, and now physicists have produced an atomic-level spin glass that can store and recall memories. The new spin glass can retrieve complete memories from partial or degraded information, acting as an associative memory. It demonstrated memory recall storage capacity well beyond what was theorized for spin glass. While the study describes proof-of-principle work done with extremely cold atoms, the system could eventually point toward more efficient AI training and resource utilization.