Artificial intelligence (AI) is transforming healthcare and is increasingly applied across the cancer care continuum. However, prior studies of its use have focused on relatively narrow applications. This scoping review characterizes the current landscape of AI applications in oncology published over the past five years. A two-part literature search was conducted in November 2024. First, nine high-impact journals were manually reviewed for articles published from November 2019 to November 2024 containing the term “artificial intelligence.” Second, a structured PubMed search using “artificial intelligence” AND “cancer” was performed, limited to the same period and filtered by study type, to classify studies by their primary area of application. Of 1,305 records screened, 575 studies met the inclusion criteria. Most studies spanned multiple medical subspecialties (n = 154, 27%) or focused on cancer detection (n = 145, 25%). The most frequently studied tumor categories were multiple tumor types (16%), breast (14%), genitourinary (13%), lower gastrointestinal cancers (12%), and lung (12%). Radiomics (n = 153, 27%) and unstructured clinical data (n = 123, 21%) were the most common data sources. The most common primary applications were predictive modeling (n = 157, 27.3%) and biomarker development (n = 156, 27.1%). Current AI research in oncology is dominated by a focus on cancer therapy and diagnosis. Future research requires prospective, multi-institutional validation and evaluation frameworks that incorporate factors such as clinical impact, workflow integration, cost-effectiveness, and ethics. Collaboration among patients, clinicians, ethicists, and healthcare leaders will be key to prioritizing research into underrepresented cancer types, mitigating AI bias, and ensuring that AI innovations are clinically meaningful and sustainable.