AI is transforming a 94-year-old Depression-era wedding dress into an interactive exhibit at the Metropolitan Museum of Art.
Monday’s Met Gala kicked off the museum’s annual Costume Institute exhibition, which focuses on “sleeping beauties” or clothes that are now extremely fragile and can no longer be worn. The exhibition features over 200 garments and accessories across 400 years and invites visitors to touch embroidered walls and experience what it was like to wear storied pieces of clothing.
But it’s the final item in the exhibition, a wedding dress designed by Callot Soeurs that New York socialite Natalie Potter wore on her wedding day on December 4, 1930, that has people — and a persona — talking. Here, the Met collaborated with ChatGPT-maker OpenAI to create a custom chatbot modeled after Potter’s personality.
The AI bot can answer visitors’ questions about Potter’s wedding, her life, and her dress — all in her persona.
Visitors just have to scan a QR code to talk to the Potter chatbot through text.
Wedding dress worn by Natalie Potter nearly 94 years ago. Credit: The Metropolitan Museum of Art
This is the first AI-aided exhibit created by the Met, with the museum’s director Max Hollein telling The Wall Street Journal that he sees the AI as a pilot program; visitor response will tell the Met more about how to further use AI.
Related: This Free Tool ‘Poisons’ AI Models to Prevent Them From Stealing Your Work — But Some Say It’s Akin to ‘Illegal’ Hacking
“I think artists will use AI in the future in very interesting and intelligent ways,” Hollein told the publication.
OpenAI trained the Potter chatbot on letters she wrote, newspaper articles, and documents from the time. According to FamilySearch, Potter passed away more than 26 years ago.
The custom chatbot with Potter’s persona was also a first for OpenAI, which says it looks for ways to collaborate with industries on real-world use cases.
“I think we have an opportunity here to do something different, and the outcome is not preordained,” OpenAI Chief Technology Officer Mira Murati told the WSJ.
Related: OpenAI Reportedly Used More Than a Million Hours of YouTube Videos to Train Its Latest AI Model
Finding a strong use case for AI is important as OpenAI faces lawsuits from creatives and pushback on copyright grounds.
Authors like Paul Tremblay and Sarah Silverman have alleged that their books were part of datasets used to train AI without their consent and artists like Billie Eilish and Jon Bon Jovi recently signed an open letter about the “catastrophic” use of AI in the music industry.
In April, the New York Times reported that OpenAI may have trained AI models on YouTube video transcriptions.
Murati spoke with the WSJ’s Joanna Stern in March and said that the company used publicly available and licensed data to train its chatbots.
Related: Tennessee Just Passed a New Law to Protect Musicians From a Growing AI Threat