Abstract
Promotion has always been a problem of attention: a firm must decide what to say, to whom, and at which moment. Artificial intelligence now participates in each of those decisions. This paper reviews how machine learning, generative models, and predictive systems have been built into promotional practice, and what scholarly and trade evidence reveals about their results. It organizes the technology into five working capabilities, follows them through six commercial settings, and examines four cases drawn from music streaming, food manufacturing, financial services, and ride hailing. It then considers how returns are measured, why outcomes differ so sharply between firms, and which risks deserve closest attention: privacy erosion, skewed ad delivery, undisclosed synthetic media, sameness of creative output, and upheaval in marketing careers. The central argument is that AI produces its strongest promotional results when it is treated as an organizational capability, supported by sound data, clear accountability, and human review, rather than as software that can simply be purchased and switched on.
