Army general says he’s using AI to improve “decision-making” (arstechnica.com)

Maj. Gen. William “Hank” Taylor, commander of the U.S. Eighth Army in South Korea, told reporters at the Association of the U.S. Army conference that he and his unit are “regularly using” an AI chatbot to modernize predictive analysis, logistical planning and routine paperwork such as weekly reports. Taylor said he’s also experimenting with building models to improve individual decision-making — helping soldiers reason through personal and organizational choices that affect readiness — noting that “Chat and I are really close lately.” The disclosure follows OpenAI data showing roughly 15% of work-focused ChatGPT conversations center on decision‑making and problem solving. The announcement is notable because it signals high-level, operational adoption of large language models (LLMs) inside military workflows, not just pilots or back‑office automation. Practically, that means LLMs are being used for forecasting, scenario synthesis and decision-support model building, which can speed planning and cognitive workloads. At the same time it raises technical and governance questions: the specific models and data handling are unspecified, leaving concerns about hallucinations, fabricated citations, model brittleness, data security, and over-reliance without robust human-in-the-loop checks. The move underscores an urgent need for validation, adversarial testing, and clear policies governing model provenance and use in mission-critical contexts.