Neural style transfer was the first widely-used "AI applied to images" technique, popularized around 2015. It uses a convolutional neural network to separate the style features of one image (a Van Gogh painting, a movie still, a fabric photo) from the content of another (your photograph) and recombine them.
For interior design, style transfer was the precursor to today's diffusion-based tools. It could give your living room photo the color and texture of a Scandinavian magazine spread, but it couldn't change the actual furniture or layout — it only restyled the existing pixels. That made it useful for moodboards and color exploration, but limited for serious redesign work.
Diffusion models have largely replaced style transfer for new tools because they handle structure as well as style. Style transfer survives in niche use — quickly remixing a render with a specific artistic look — but it's not the right tool for "show me this room as a different design".