[ad_1]
However, some experts warn of the real cost of free and open generative AI tools: although they may be more democratic in terms of access, the image database will be deleted from the entire Internet, so the data will not be provided voluntarily. “Laws and regulations around AI need to evolve as quickly as the technology itself,” Noel says.
The problem is that these paid programs are not accessible to marginalized groups. Jennifer Lee, an assistant professor at the Fashion Institute of Technology, says less than 10 percent of her 3D apparel visualization classes are made up of minority students due to financial constraints, which is why she volunteered to teach the program to a minority group for free in 2021. Students. He said that funding from institutions and brands is needed to give marginalized people a seat at the AI table – which brands will benefit because behind-the-scenes differences can reduce some of the misrepresentation.
AI training for better representation
Many inclusive brands relish the opportunity to participate in a solution to undo biases built into algorithms. “If you look at the millions of images on the internet, you’d never know that plus-size women are largely invisible,” says Nadia Bujarwah, CEO of plus-size specialty retailer Dia & Co. To remove all the biases that are deeply ingrained in what we do, but if we consciously reverse those biases, that’s the most exciting outcome of all. However, the brand is taking a cautious approach given the controversy and has yet to test image-based AI.
The image-based generative AI that underwear brand Adore Me tried isn’t advanced enough to accurately represent items like bras and corsets, but the brand created its own big-language AI models for text-based tasks, which VP of strategy Ranjan Roy says was a much simpler process. Roy thinks brands looking to make sure their image-based AI makes a difference should look to build their own custom models and datasets, because AI can go in and highlight the same biases in human-generated images on the internet.
“The data sets that will drive business in the next five to 10 years are being built right now, and the training sets we’re using now are biased or built with different populations. It’s really exciting and important for brands to be able to train image models with different models instead of using AI to replace them.”
Brands can use generative AI as a way to create test campaigns and gather feedback on campaign concepts from consumers and employees, said Ramona Dunlap, assistant professor at the Fashion Institute of Technology. They then take those insights and feedback and work with agencies and photographers to create an optimized version of the campaign that really speaks to people from different communities.
Brands that want to succeed in accelerating AI initiatives need to deeply understand what their audience really wants and values. “Too often in companies we might sit at a desk and think we know what’s best and not put it in the hands of consumers — and that’s when they get a reaction,” says Dunlap. “This makes the company spend twice as much money, because they’re going out there and investing in something that their audience will never want.”
Comments, questions or comments? Email us. feedback@voguebusiness.com.
More from this author:
Ferragamo’s sales fell by 4% during the ‘transition period’
Gucci among fashion companies targeted in EU antitrust raids
Fashion can now report its use of plastic. Does reduction follow?
[ad_2]
Source link
