Intelligent Voice Assistants (IVAs) have become integral to many users' daily lives, using advanced algorithms to automate various tasks. Nevertheless, many users do not understand the underlying algorithms and how they work, posing potential risks to the competent and self-determined use of IVAs. This work develops three online training modules to promote algorithm awareness, providing (1) basic knowledge of algorithms, (2) risks posed by algorithms in IVAs, and (3) scientific evidence on algorithm aversion. A total of 110 participants were studied to analyze the training effects on various perception levels relevant to IVAs, including usage perception (attitude, exploration), privacy aspects (trustworthiness, privacy control), persuasion aspects (persuasion knowledge, anthropomorphic perception), and self-determined interaction variables (reflection, indulgence). The results show that the modules increase awareness of IVA algorithms, influence user perceptions (e.g., higher exploration intentions), and promote critical engagement with IVAs (e.g., lower trustworthiness). Moreover, the modules contribute to a higher sense of privacy control, reduce persuasive perceptions of IVAs (e.g., anthropomorphic perception), and promote self-determined interaction (e.g., higher indulgent use). The modules offer a new approach to promoting the competent use of IVAs in society and provide a starting point for further research and educational institutions to increase algorithm awareness for IVAs and other AI-based systems.