The popularity of artificial intelligence (AI) and its applications has been steadily growing across various disciplines, including the field of science. In life sciences, AI has been increasingly applied for imaging flow cytometry towards automated management and classification of biological cell image data. However, when considering imaging flow cytometry for cell sorting, the need for employing AI over more facile, traditional image analysis methods are not as pronounced. This is primarily due to the time restrictions between image acquisition and sorting actuation that inevitably come with cell sorting. AI-enabled image-activated cell sorting (IACS) methods remain limited, and recent advancements in IACS still find success while relying on traditional feature gating strategies. Here, we compare the performance of feature gating, classical machine learning (ML), and deep learning (DL) in the differentiation of Saccharomyces cerevisiae mutant images taken by our intelligent IACS system to assess the advantages of AI for image classification in IACS. We show that while both classical ML and DL impart an increase in enrichment capability over feature gating, employing classical ML resulted in a smaller improvement at a larger cost of longer processing time than DL. We further performed IACS on mixed mutant populations and quantified target enrichment via downstream DNA sequencing to confirm the applicability of DL for the proposed study. Our findings validate the benefit and practicability of employing DL in IACS for microscopy-based genetic screening of S. cerevisiae, encouraging its inclusion for future advancements in this field.