As the aging population continues to grow, understanding older adults’ preferences for artificial intelligence (AI) versus human assistance in daily tasks is crucial for developing effective AI to support healthy and active aging. This study aims to identify patterns in older adults’ task assistance preferences, and variations based on demographics, self-reported health, prior AI attitudes, and AI experience. A cross-sectional questionnaire survey was conducted with 315 older participants who rated 49 tasks for human versus AI assistance. Results revealed a slight overall preference for AI, with significant variation across tasks. Older adults favored AI for tasks that were cognitive and functional in nature, such as giving directions and information gathering, while human assistance was preferred for tasks tied to emotional or social interactions, such as feeding or emotional companionship. These preference patterns were consistent across age, education, self-reported health status, and prior AI experience. Subgroup analyses indicated that men, those with positive attitudes toward AI, and those married but living apart showed stronger preferences for AI, particularly for routine tasks. These findings provide insights into older adults’ task assistance preferences and can guide the design of AI tailored to their needs, emphasizing the variability in preferences across task types and individual variations.