Generative artificial intelligence for nanomaterial discovery and nanoengineering: From data-driven design to autonomous laboratories

From conventional trial-and-error experimentation, nanomaterials research is moving toward data-centric and autonomous discovery paradigms that couple nanotechnology with artificial intelligence (AI). In this review, we critically examine how generative artificial intelligence, including generative adversarial networks, variational autoencoders, diffusion models, transformer architectures and large language models, enables inverse design of nanomaterials and supports nanoengineering across diverse application domains. The review provides a structured and critical survey on AI-driven nanomaterial discovery, focusing on studies where generative models explore nanoscale composition–structure spaces and where self-driving laboratories (SDLs) close the loop between design, synthesis, characterization and performance evaluation. We provide a comparative analysis of major generative architectures for nanomaterial tasks, highlighting differences in validity, novelty, diversity, conditional control and computational cost. Representative case studies are discussed in which generative models and autonomous experimentation have led to experimentally validated nanomaterials, such as quantum dots, nano-porous frameworks and functional nanocomposites. A dedicated section analyzes data infrastructures and nano-specific challenges, including size-dependent phenomena, surface and interface dominance, defect sensitivity, nano-bio interactions and dataset bias. Fundamental limitations of artificial intelligence in nanomaterials are examined, covering data scarcity for morphology-rich systems, interpretability, reproducibility, safety and regulatory aspects. Finally, the review outlines research priorities for multimodal foundation models, physics-informed generative frameworks, agentic AI scientists, digital twins (DTs) and next-generation SDLs that can accelerate nanomaterial discovery while remaining scientifically reliable and ethically responsible. Together, these perspectives position generative AI as a key engine for future nanoengineering, provided that nano-specific data ecosystems and rigorous validation workflows are developed alongside algorithmic advances.