Think you can spot a ChatGPT image? Take my real-or-AI quiz and see how many fool you

OpenAI recently released ChatGPT Images 2.5, introducing a range of enhancements to the image generation process. This new model boasts sharper details, quicker generation times, more accurate editing options, and tools that facilitate image creation and sharing. Among the key improvements are features like Sketch, which transforms rough sketches into polished images, templates for posters, and advanced editing tools.

However, despite these advancements, there are still concerns surrounding the presence of noticeable noise patterns in certain AI-generated images. TechRadar's Eric Hal Schwartz highlighted this issue, noting that the images often exhibit a hazy, sandy appearance, making them less clean and polished overall.

While the latest version of ChatGPT Images has undoubtedly improved, the question remains whether it has reached a point where AI-generated images can convincingly mimic genuine photographs. To explore this matter further, I conducted an experiment using both real images from my summer travels and AI-generated images, aiming to create a fair comparison.

For each subject, I asked ChatGPT to produce images with similar compositions and settings to those I captured during my trips. The goal was to generate images that were as ordinary and unremarkable as possible, rather than producing striking or artistic AI visuals.

Now, I invite you to take the following quiz to determine which images are real and which are the result of AI creation. Are the familiar "gritty, sandy" effects still apparent in the AI-generated pictures, or do we find ourselves unable to discern the differences between genuine and artificial images? Share your results in the comments below and join the discussion on the evolving capabilities of AI image generation.