Generative AI is beginning to reshape not just business operations but the way companies think, according to a new analysis in the Harvard Business Review. While many executives focus on efficiency and cost reduction, the authors argue that the technology’s greater promise lies in the ideation phase—the early stage of innovation where assumptions are challenged and new possibilities emerge.
The HBR piece notes a sharp rise in companies using large language models for idea generation. In a recent survey, that use case jumped 12% from 2024 to 2025, entering the top five most common applications.
Large language models, or LLMs, can rapidly produce hundreds of variations on a concept, helping teams iterate more deeply, and their exposure to vast datasets allows them to remix insights across industries in ways that humans may not naturally attempt.
But the researchers caution that this creative boost comes with limitations. Because AI models draw on patterns in existing data, their ideas can default to the familiar unless users apply targeted techniques—such as persona-based prompts, hybrid queries, or step-by-step reasoning—to push them toward more unconventional thinking.
Without that discipline, the authors write, AI-generated ideas may start to look increasingly alike, even as teams feel more productive.
The article also highlights new models of human–AI collaboration. In some settings, AI leads by generating initial concepts and refining language; in others, it serves as a thought partner, probing assumptions or mimicking customer reactions. The authors note that flexibility is already influencing marketing, product development and strategic planning.
