Artificial Intelligence

Flaws in large language model-based user simulators have been exposed by research from Google Research introducing ConvApparel, a dataset of more than 4,000 human-AI conversations in the apparel shopping domain. The work develops a three-pillar evaluation framework to measure and address the realism gap in AI testing, with data-driven simulators outperforming prompt-based approaches across key behavioural metrics.

Fashion retail is moving into a sharper phase of algorithmic pricing as AI shopping tools connect consumer preferences, inventory pressures and automated purchasing. Dynamic pricing, virtual try-ons and agentic commerce could make shopping easier, but may also deepen discount chasing, impulse buying and overconsumption in already high-volume markets shaped by markdowns.

IIT Roorkee has developed a new artificial intelligence framework that enables the translation of detailed textual descriptions into professional-quality fashion sketches. Built around the FLORA dataset and the NeRA adapter, the research addresses limitations in existing AI tools and supports more precise, expressive and sustainable design workflows across fashion, retail and allied creative industries.

Researchers at Pusan National University have developed a systematic prompting method to use ChatGPT for predicting fashion trends. Their Top-Down Prompting technique, inspired by the Lotus Blossom approach, helps generate more accurate and consistent insights, potentially democratising trend forecasting for smaller players and fashion education.

Researchers in China have used artificial intelligence to design better carbon-based materials that can clean toxic dyes from industrial wastewater. By analysing hundreds of lab results, the system identified the most effective combinations for dye removal, helping make low-cost, eco-friendly solutions for the textile and chemical industries more efficient.

Artificial intelligence is reshaping agriculture by offering smarter ways to monitor pests and optimise pesticide use. A new initiative demonstrates how rural communities can harness cutting-edge technologies to strengthen farming practices, protect health and build innovation capacity. The project also highlights how collaborative partnerships can bring long-term economic and community benefits outside urban centres.

Korean researchers have demonstrated that generative AI can successfully predict and visualise seasonal fashion trends, achieving accurate implementation in over two-thirds of tested cases. The Pusan National University study used ChatGPT and DALL-E to forecast Fall/Winter 2024 menswear collections, generating images that closely matched real runway designs whilst highlighting the continued importance of fashion expertise.

Virtual influencers — digitally created personas used to provide entertainment, generate content and endorse brands — are becoming increasingly influential, especially among Gen Z and digital-first audiences. The rise of virtual influencers is notably prevalent in Asia.

A truly global phenomenon, the transformative power of artificial intelligence or AI can reach its full potential only when paired with human expertise and ingenuity. A KPMG study on the apparel and footwear industry finds that although it is being deployed, the journey still has miles to go.

The National Institute of Fashion Technology (NIFT) has launched the ‘VisioNxt Fashion Forecasting Initiative’, a bilingual web portal and the India-specific fashion trend book ‘Paridhi 24x25’.