Yesterday we demonstrated the use of ChatGPT in a fully autonomous counter-messaging pipeline, designed to automatically monitor television news coverage from Iran in Persian, identify statements criticizing the JCPOA, craft counter-messages written for different audiences (a global audience versus a domestic Iranian audience) in different languages (English and Persian) and platforms (Facebook, Twitter, LinkedIn, Telegram and news outlets), along with creating an associated image prompt for each message that could be used with any image generation platform.

To explore what the accompanying images might look like, we pasted the ChatGPT-generated image prompts into Midjourney v5. We've updated the original post with the generated images and you can also see them below in order. Despite v5's photorealistic results, there are still imperfections in each, subtle and not-so-subtle that give away each as being machine-generated, but despite these limitations, the results are nothing short of stunning, showcasing just how far image generation technology has come and just how potent it already is.