AI anthropomorphism and its effect on users' self-congruence and self–AI integration: A theoretical framework and research agenda