Table 1. Summary of generative AI models applicable to meat processing

Model type	Core mechanism	Strengths	Applications in meat processing	References
Generative adversarial networks (GANs)	Adversarial training between generator and discriminator	High-resolution image generation; defect simulation	Simulating meat surface defects for data augmentation and model training	Fu et al. (2023) Goodfellow et al. (2014) Lu et al. (2022)
Variational autoencoders (VAEs)	Latent space encoding with probabilistic reconstruction	Effective in simulating sensory attributes and structural variance	Generating synthetic quality profiles and 3D meat texture representations	Kench and Cooper (2021) Kingma and Welling (2019) Zhang et al. (2024b)
Large language models (LLMs)	Transformer-based sequence modeling	Automated report generation; summarization of unstructured data	Predicting quality trends from process records; real-time documentation	Brown et al. (2020) Li et al. (2024b) Raza et al. (2025)
Multimodal large language models (MLLMs)	Integration of textual, visual, and sensor data using transformers	Cross-modal learning; data fusion and flexible decision support	Real-time interpretation of sensor/image/text data in smart factories	Han et al. (2025) Piechocki et al. (2023) Zhang et al. (2024a)