Look for Physical and Logical Flaws
The classic advice to 'count the fingers' is becoming less reliable as AI models improve, but anatomical and physical errors remain a key tell. Look for hands and fingers that seem to bend unnaturally, merge together, or have incorrect proportions. Eyes
can also be a giveaway; check for pupils that aren't perfectly round, strange light reflections, or a vacant, lifeless stare. Beyond people, inspect the physics of the scene. Are shadows falling in a consistent direction from the light source? Do reflections in windows or water accurately mirror their surroundings? AI often struggles with these logical details, creating scenes where things just feel slightly 'off'. Look for repeating patterns in textures like skin or backgrounds, which can sometimes look unnaturally uniform, like they were digitally stamped.
Analyse Text and Fine Details
One of the most persistent weaknesses in AI image generation is text. If an image contains signs, book covers, or labels, zoom in. AI-generated text is often a garbled mess of characters that look like letters but don't form real words. This applies to any script or symbol system. Even if other parts of the image look flawless, nonsensical text is a huge red flag. Other fine details can also reveal an AI's hand. Look for jewellery that seems to float slightly off the skin, glasses that lack earpieces, or parts of a subject that blend confusingly into the background. These minor errors break the illusion of reality.
Consider the 'Too Perfect' Problem
Sometimes, the clue isn't an error but an unnatural perfection. Many AI images have an 'uncanny valley' quality where everything is too smooth, too symmetrical, or too aesthetically pleasing. Real-life photos have imperfections: slight motion blur, lens distortion, uneven lighting, and natural human asymmetry. AI-generated portraits, in contrast, often feature skin that's as smooth as plastic, with no visible pores or minor blemishes. The composition might feel too balanced, and the colours overly saturated, giving it a synthetic, hyper-real feeling that separates it from authentic photography or human-made art.
Look Beyond the Image: Provenance and Watermarking
Instead of just trusting your eyes, do some detective work. One of the most powerful tools is a reverse image search using services like Google Lens. This can help you trace the image back to its original source. If it first appeared on a site known for digital art or was posted by a self-described AI artist, you have your answer. Increasingly, tech companies are embedding invisible watermarks or metadata into AI-generated content. Standards like Content Credentials (C2PA) create a verifiable record of an image's origin and edit history. While not foolproof, checking an image's metadata or using verification tools can provide strong evidence of AI involvement.
The Crucial Step: Don't Jump to Accusations
Spotting a potential AI tell is one thing; accusing a creator is another. The goal of media literacy is not to start a witch hunt. Before you comment 'This is AI!', consider the context. Is the creator a photographer known for their real-world shots, or a digital artist who often experiments with new tools? Many artists now use AI as one part of a larger creative process, blending it with their own skills. Accusing them of 'faking it' misunderstands the nature of their work. Furthermore, many creators are transparent about their methods if you check their post descriptions, bios, or other work. A lack of disclosure isn't an automatic sign of deception.
How to Ask a Creator (Politely)
If your curiosity is overwhelming, there's a respectful way to ask. Instead of a blunt 'Is this AI?', frame your question with appreciation for their work. Try something like, 'This is a fascinating image! I'm really interested in your creative process. What tools or techniques did you use to achieve this look?' This approach shows genuine interest rather than suspicion. It opens the door for a conversation about their artistry, which may or may not include AI. Remember that creators are often proud of the tools they master, and many are happy to share their process when asked respectfully. This fosters a more positive and educational online community for everyone.
