Researchers at OpenAI have found that large transformer models can generate coherent images when trained on pixel sequences. They compared the quality of generated images to image classification accuracy and found that their best model is competitive with top convolutional neural networks in unsupervised settings. This suggests a potential link between generative capabilities and image recognition abilities. --- Why it matters: This matters because it could lead to more efficient and effective training of AI models for image generation and recognition tasks, potentially reducing the need for large datasets and computational resources. Source: https://openai.com/index/image-gpt