Artificial Intelligence in drug discovery and development: current landscape, challenges, and future perspectives.

Artificial intelligence (AI) is transforming drug discovery and development, fields historically constrained by long timelines, high costs, and substantial attrition. Recent advances, particularly in generative modeling, enable an accelerated and increasingly systematic exploration of vast chemical and biological spaces, improving molecular interaction modeling and streamlining the identification and optimization of therapeutic candidates. However, the true utility of this expanded search space remains strictly bounded by the quality of upstream data and the logistical constraints of downstream experimental validation. Emerging platforms, including scaffold-aware and 3D molecular design tools (e.g., AlphaFold, MoleR, and PocketCrafter), single-cell foundation models, and large language models (LLMs), are expanding AI's applicability across the research and development pipeline, spanning target identification, drug discovery, lead optimization, phenotypic screening, and precision biology.AI is also increasingly integrated into preclinical and clinical research workflows, informing adaptive trial design, enabling AI-driven drug repurposing, and supporting the development of safer and more personalized therapies. While the U.S. FDA has approved numerous AI-enabled medical devices and software tools, no fully AI-discovered and AI-designed drug has yet received marketing approval. Nonetheless, several AI-originated candidates have progressed into clinical development, underscoring AI's growing translational impact. Collectively, these advances position AI as a collaborative "lab partner," capable of uncovering non-intuitive molecular designs, accelerating target and lead optimization, and enabling exploration of previously inaccessible chemical and biological space to inform downstream development and clinical decision-making. Despite gains in efficiency, scalability, and cost reduction, the broader impact of AI depends on access to high-quality multimodal data, robust regulatory and ethical frameworks, and careful recognition of methodological limitations. This review critically examines the evolution of AI approaches, highlighting key challenges and opportunities that shape the future of data-driven therapeutic innovation.