Can Generic LLMs Truly Understand Fashion Trends and Provide Actionable Insights?

In a recent Vogue test, the publication set out to determine whether generative AI chatbots could grasp the nuanced world of fashion and provide relevant styling advice. The experiment involved comparing AI assistants from major tech companies, including Google, Meta, and Microsoft, with the evaluation of Vogue editors, who tested the bots' ability to understand fashion trends and style. The big question: can generative AI, which excels at aggregating facts, offer actionable insights into a domain as subjective and fluid as fashion?

Three AI assistants — Claude, Meta AI, and Copilot — were tasked with identifying summer fashion trends. The results, reviewed by Vogue's senior market editor, Madeline Fass, revealed both hits and misses, with many recommendations being either outdated or lacking the depth required for meaningful styling advice. The test revealed significant limitations in these AI models’ abilities to understand trends in a practical and culturally relevant way.

The Limitations of General AI Models for Fashion

The test demonstrated the current limitations of general AI models in fashion. Claude, an AI assistant trained on data up until August 2023, was asked, "What is in style this summer?" It listed neon colors, retro prints, cargo pants, mini bags, and chunky sandals. While some trends, like cargo pants and mini bags, were spot on, neon colors were outdated — a trend that peaked five years ago. This issue highlights the fact that general AI models often rely on outdated or incomplete data, making them less capable of offering relevant, real-time insights in a fast-evolving industry like fashion.

Similarly, Meta AI, which sourced its information from Google, fared slightly better, recommending oversized shirts and wide-leg trousers. However, it also included a suggestion for a faux leather jacket, which was impractical for summer. Copilot, the third assistant tested, took a more detailed and poetic approach, referencing trends like “butter yellow” and “rose details,” but failed to provide a clear reasoning behind its choices.

As Christopher Wylie, founder of fashion AI consultancy Supermodel, noted: “Once again, my feeling was that the answers were decidedly plain. It’s very basic because a lot of these models aren’t trained specifically on fashion data. It has a basic knowledge on a superficial level, but you cannot use those to necessarily recommend or design an outfit.” This perfectly sums up the core issue with these general models: they lack the specialized knowledge needed to navigate the intricacies of fashion.

The Need for Fashion-Domain Multimodal LLMs

As the Vogue test demonstrated, current AI models are not well-equipped to handle the nuances of fashion, which go beyond simply recognizing garments or colors. Fashion trends are deeply tied to cultural, emotional, and aesthetic shifts — areas that general AI models are not trained to interpret.

This is where fashion-specific AI models, like Trendee's fashion-domain multimodal large language model (LLM), come into play. Unlike general models, Trendee’s LLM is trained on fashion-specific data from a wide variety of sources, including text, images, and videos. This allows it to better understand not just what trends are emerging, but the cultural and emotional context driving those trends.

For instance, while general models can suggest basic trends like wide-leg trousers or mini bags, they struggle to recognize micro-trends that emerge on social platforms like TikTok or Instagram, such as balletcore or mermaidcore. Trendee’s multimodal LLM, by analyzing fashion across platforms in real time, can spot these trends early and offer designers actionable insights, helping them stay ahead in a highly competitive industry.

Making Trends Actionable: From Insight to Design

One of the biggest challenges in fashion trend forecasting is not just identifying trends, but translating them into actionable insights for design and product development. While the AI assistants in Vogue’s test could list general trends, none provided guidance on how to implement these insights into practical design choices. For example, knowing that cargo pants are trending is helpful, but without detailed insights into fabric preferences, color palettes, or styling suggestions, the information remains superficial.

Trendee’s approach solves this issue by providing more than just trend identification. It offers in-depth analytics on consumer preferences and cultural shifts, enabling fashion designers to make informed decisions on product development. By understanding the deeper context behind a trend, such as why a particular style resonates with a certain demographic or how it ties into broader lifestyle changes, designers can create collections that not only reflect current trends but also anticipate future shifts.

Challenges in AI-Generated Design

Another limitation highlighted by the Vogue test is that while AI can identify trends, it struggles with design practicality. Many generative AI models can create imaginative, visually striking designs, but these designs often lack feasibility when it comes to actual production. AI-generated designs may be too artistic or abstract, ignoring the practical considerations that human designers take into account, such as fabric availability, manufacturing techniques, and consumer wearability.

While AI can certainly inspire creative ideas, it cannot replace the human touch that is essential in fashion design. Designers bring an understanding of materials, craftsmanship, and marketability that AI, at least for now, cannot replicate.

Conclusion

The Vogue test illustrates both the potential and limitations of generative AI in fashion. While AI assistants can provide a surface-level understanding of trends, they lack the cultural and emotional depth needed to truly excel in this space. Fashion is about more than just clothing; it’s about understanding lifestyles, aesthetics, and the rapidly changing tastes of consumers.

General AI models simply don’t have the specialized training required to navigate this complex industry. That’s why the future of fashion forecasting lies in fashion-domain multimodal LLMs, like Trendee's, which integrate data from multiple sources and offer actionable insights. These models, combined with human expertise, will be key to staying ahead in the fast-paced world of fashion, making AI a powerful tool for designers, brands, and retailers alike.