Generative artificial intelligence includes models that approximate data distributions and can generate new, plausible data points.
Key methods include Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and autoregressive models such as GPT.
These models learn by optimizing loss functions aimed at making the generated data indistinguishable from real training data.
Generative AI has significant applications in image synthesis, natural language processing, and simulation of complex systems.
Definition:
“Generative artificial intelligence refers to systems capable of generating new content by learning the underlying patterns of a given dataset, producing outputs that resemble the original data distribution.”
Source:
Goodfellow et al. – Generative Adversarial Nets, 2014; as well as current review articles on generative AI.