Palliative care is an interdisciplinary approach to specialized care focused on improving quality of life for patients with serious illness. Once concentrated on end-of-life, the field has evolved to address multidimensional needs across multiple disease trajectories, care settings, and patient populations. Artificial intelligence/machine learning (AI/ML) algorithms represent a promising approach to scaling this expanded vision by identifying patients with unmet palliative needs earlier and more consistently than traditional screening methods. This chapter examines AI/ML applications across three algorithmic domains: mortality prediction, modeling specific palliative needs such as symptom burden, and modeling goal-concordant care. We trace a conceptual progression from “Who will die soon?” to “Who is suffering now?” to “Is this patient receiving care aligned with what matters most to them?” and argue that the field's concentration on mortality prediction reflects technical accessibility rather than alignment with palliative care's broadened scope. Despite this mismatch, emerging work in symptom modeling and preference extraction demonstrates movement toward more clinically meaningful targets. Finally, we examine the growing body of implementation and evaluation studies, which represent the definitive test of whether these algorithms improve care delivery. Findings show improvements in process measures such as consultation rates, but evidence of patient-centered outcomes remains sparse, and challenges around model trust, generalizability, and long-term monitoring persist. Addressing these gaps will require reorienting algorithmic targets toward contemporary palliative needs, equitable deployment across diverse settings, and rigorous evaluation of outcomes that matter to patients and families.