This paper explores how artists use machine learning (ML) processes to create ambiguity in AI art, challenging traditional HCI views on ML as a design material. The authors analyze nine AI artworks that use computer vision and image synthesis, revealing how artists engage with the ML process (dataset curation, model training, and application) to evoke ambiguity. They argue that ML should be conceptualized as design elements rather than technical details, emphasizing the importance of process-centered design in HCI. The study highlights how ambiguity in AI art challenges established desiderata like dependability and explainability, offering alternative approaches to ML in HCI. The authors propose that ambiguity of process is a new type of ambiguity relevant to ML experiences, suggesting that HCI should shift focus from artifacts to the processes that create them. The paper also discusses the role of uncertainty in ML, arguing that it is inherent to the process and cannot be fully eliminated. The findings suggest that designers should work more closely with technical teams to understand the ML pipeline and its implications for user experience. The study contributes to the broader discourse on AI art, emphasizing the creative potential of ML processes in generating ambiguity and challenging conventional assumptions about ML in HCI.This paper explores how artists use machine learning (ML) processes to create ambiguity in AI art, challenging traditional HCI views on ML as a design material. The authors analyze nine AI artworks that use computer vision and image synthesis, revealing how artists engage with the ML process (dataset curation, model training, and application) to evoke ambiguity. They argue that ML should be conceptualized as design elements rather than technical details, emphasizing the importance of process-centered design in HCI. The study highlights how ambiguity in AI art challenges established desiderata like dependability and explainability, offering alternative approaches to ML in HCI. The authors propose that ambiguity of process is a new type of ambiguity relevant to ML experiences, suggesting that HCI should shift focus from artifacts to the processes that create them. The paper also discusses the role of uncertainty in ML, arguing that it is inherent to the process and cannot be fully eliminated. The findings suggest that designers should work more closely with technical teams to understand the ML pipeline and its implications for user experience. The study contributes to the broader discourse on AI art, emphasizing the creative potential of ML processes in generating ambiguity and challenging conventional assumptions about ML in HCI.