Recent advancements in artificial intelligence (AI) and digital imaging technologies have enabled the automation of pigmented skin lesion counts, yet the confidence in and attitudes towards automated counts among researchers and clinicians remain uncertain. This Delphi consensus study aimed to establish a standardised protocol for pigmented lesion counting. It involved 25 internationally renowned skin cancer and melanoma clinicians and researchers, and we obtained their perceptions of using AI and digital imaging for lesion counting. 64% of the experts had prior experience with AI tools or were familiar with lesion counting using AI in research studies. There was no consensus among the experts regarding whether AI is more accurate (31% agreement) or reproducible (46% agreement) than human observers in pigmented lesion counting. Only 42% of the experts expressed confidence in employing AI for pigmented lesion counting in research studies. While AI’s current state may be perceived as unreliable for some individuals (e.g., highly sun-damaged skin or many seborrheic keratosis), the experts emphasized that the scope of automated counting should not be restricted to only certain groups, as its accuracy is rapidly improving. Recommendations to increase sensitivity and specificity of AI included using carefully curated, more inclusive training datasets, using standard protocols, and considering a broader range of labelled lesions. Validation studies across population subgroups were also deemed essential. These insights will contribute to developing more effective AI solutions for automating counts of pigmented lesions, ultimately aiding in better risk stratification for skin cancers.