This theoretical article approaches the impact of generative artificial intelligence in music not merely as “a new tool for composition/production”, but as a multiplier that reconfigures—both in speed and in scale—a sociotechnical regime of selection operating across the tool–interface relationship, platform infrastructures, listener practices, and legal/regulatory frameworks. The text distinguishes end-to-end automation (where, with a single command, decisions regarding form, harmony, arrangement, performance, and post-production are generated largely by the model) from partial/auxiliary task automation and argues that prompting-centred workflows redistribute expertise, time, and responsibility. From the perspective of complexity science, it is proposed that musical meaning and creative value are not “fixed” qualities embedded within the work, but rather emerge within processes of emergence (the derivation of higher-level patterns from lower-level interactions), feedback, and path dependence (the accumulation of small early differences into long-term inequalities), through interactions between production practices and platform circulation. The study further discusses how listening contexts shaped by provenance and labelling information may reframe listeners’ judgements via the effort heuristic (the inference that ’more effort = higher value’), and how, in content economies with low reproduction costs, trust may be re-established through costly commitments (costly commitments: hard-to-imitate trust signals involving labour/risk). In conclusion, the article suggests that keeping the unit of analysis in musicology solely at the level of the work is insufficient; instead, there is a need for mixed-method research programmes that relate micro (perception), meso (production), and macro (platform selection/circulation) levels.