The article reports that while generative AI tools can generate lots of preliminary ideas and objectives (“brainstorming”), those AI-generated goal-lists are often incomplete, redundant, or mis-prioritized — meaning human expertise remains essential to turn those ideas into high-quality, decision-ready frameworks.
Generative AI (GenAI) models — including GPT-4o, Claude 3.7, Gemini 2.5 and Grok-2 — can produce individually plausible objectives for organizational or policy decision-making, but when judged as complete sets, these AI-generated objectives tend to be incomplete, redundant, and often include “means objectives” rather than true fundamental objectives.
Because of these structural shortcomings, the study concludes that GenAI output should not be used alone for high-quality decision analysis; human expertise is essential to refine, validate, and restructure the objectives to ensure coherence, completeness, and logical organization.
When GenAI brainstorming is combined with careful human oversight — using structured prompting (e.g. chain-of-thought) and critique-and-revise workflows — the quality of the resulting objective sets improves markedly, producing smaller, more focused, logically consistent and usable frameworks for decision analysis.
“Generative AI performs well on several criteria, but it still struggles with producing coherent and nonredundant sets of objectives. Human decision analysts are essential to refine and validate what the AI produces," says Simon.Read here.