Can AI figure out what people like, spot upcoming trends, and make brands look awesome, all by just looking at images?

I did my own unscientific test to see how well AI can analyze a brand based on a photo. Using a very short prompt and a photo I took at Whole Foods in the wine section, I submitted the same prompt to both ChatGBT 4o and Claude 3.5.

My prompt: You are a creative director at a design agency. Can you tell me about the branding of this bottle.

Within seconds I had an interesting answer from each—and a clarification from Claude when it correctly identified the bottle and then corrected me by telling me that it was NOT wine, but Sangria. 

You can read the full responses from each in the images below.

By studying tons of examples and learning from them, AI systems get really good at recognizing patterns and understanding what works well in design. AI might sound a bit like magic, but it actually uses a complex process called deep learning.

Both responses were similar but had some differences in the amount of detail and reasoning behind their answers. They covered topics like color scheme, bottle shape, logo design, font choice, labeling, target audience and overall aesthetic. ChatGBT’s response even pointed out packaging details such as: "The unique swing-top closure not only adds a vintage, artisanal feel but also emphasizes convenience and practicality, suggesting that the product is easy to open, reseal, and share. This feature enhances the brand's appeal as a shareable drink for casual gatherings.”
The analysis of the overall brand aesthetic and target audience led me to check out the Lolea website and do my own analysis of the brand’s aesthetic. See the images for a collage of the website imagery.

What do you think? Was AI able to analyze the product based on a photo I took in a grocery store? 

You can read more about this AI exploration on my LinkedIn post.