This paper empirically examines the boundary between human and machine creativity in cultural and creative industries by analysing 593 tasks across 126 creative occupations. We conceptualise creativity through an evolutionary perspective, viewing it as navigating rule ambiguity across three phases: retention (codified rules), adoption (tacit rules), and origination (novel rules). Employing synthetic data generation to measure cognitive and behavioural rule characteristics at task level, based on the December 2023 Australian Skills Classification, we identify two key mechanisms determining the creativity frontier. The structured novelty effect shows that AI autonomy increases significantly when codified cognitive rules combine with novel rule creation, indicating that machine creativity operates through efficient exploration linking retention and origination phases to support recursive innovation. The tacit knowledge boundary establishes that tacit rules create the strongest constraint on AI autonomy, preserving human efficacy advantages where experiential knowledge resists algorithmic capture. These findings show that machine creativity thrives through systematic exploration within structured boundaries, while human creativity maintains advantages in adaptive rule selection requiring contextual judgement. Creative work operates predominantly through human-AI collaboration (86.3% of tasks), with limited zones of pure substitution (2.7%) and AI immunity (11.0%). We distinguish efficacy (adaptive rule selection) from efficiency (optimised execution), advancing understanding of human-AI complementarity beyond productivity-focused frameworks. Economic value emerges from strategic coordination where humans provide adaptive efficacy while AI contributes structured exploration efficiency. For creative industries, this requires developing organisational practices leveraging both modes: investing in structured novelty capabilities while protecting tacit knowledge, and designing for human-AI collaboration.