Can AI remove bias in fashion recruitment?

To receive the Vogue Business newsletter, sign up here. When Laurent Piffaut was working as the global retail director for a luxury fashion company in 2018, crisis hit when a senior member of staff at the brand’s Madison Avenue store had to leave on short notice. “Already at capacity, we were scrambling to deal. I started getting frustrated at what seemed like long lead times and complicated recruitment processes to find someone who needed to start immediately,” says Piffaut. He saw an opportunity to create a fashion-specific talent marketplace that would quickly find candidates to fill retail roles, particularly on a freelance or short-term basis. In 2019, Piffaut — who has worked in the luxury industry for 16 years and held positions at Prada, Alexander McQueen, Dior and Chanel — joined forces with Eli Duane, a recruitment specialist who had worked in financial services, and Andreas Asprou, a software engineer who co-founded successful social marketing startup Flick at 19, to launch fashion recruitment platform Dweet. The platform has since expanded to offer a wider range of roles in fashion and luxury, from marketing to design to HR positions. Over 130 companies are on board, including Chanel, Louis Vuitton, Puig, Valentino and Lanvin. It’s a critical and dynamic time for fashion recruitment. The pandemic gave rise to new ways of working, with many leaving their jobs when companies did not provide sufficient support for remote working. By 2021, the great resignation was in full swing, lasting through 2022, with the majority of workers citing low pay and few opportunities for advancement. Companies have been scrambling to improve their hiring processes, widen and diversify the talent pool, improve retention and job satisfaction, and offer more opportunities for remote and temporary work. “Management can sometimes feel a bit traditional and hierarchical, and I’ve seen many talented people have to compensate for a lack of structure and resources,” says Piffaut. “I found myself stuck within an excessively red-tape bureaucracy, making hiring additional resources or trialling a new way of working or different types of profiles difficult. The industry is often tapping into the same pool of talent and struggling to expand its reach to new sources or more diverse talent.” Dweet’s goal is to streamline the process, allowing brands to hire talent that’s of higher quality than a traditional agency and do it quicker, cheaper and on a more flexible basis. Its founders saw the opportunity to combine the use of tech, which fashion has historically been slow to adapt to, and the human element, which fashion requires to a greater extent than other industries. The company expects revenue to reach around £4 million by the end of this year, up from £1.8 million in 2022 and is aiming to become profitable by the end of 2023. At the centre of its offer is its AI-matching algorithm, which speeds up and simplifies the way candidates are evaluated. This is balanced with the strong relationships with brand partners that Dweet builds to ensure brands’ needs are understood and met. But with opinion divided over whether AI helps to eradicate discrimination — or simply reinforces biases — can Dweet achieve the scale needed to transform the hiring process? Removing bias Dweet uses AI algorithms to select top candidates who have applied to a certain role — a method that is used often in industries such as tech or financial services but seldom in fashion, and which has provoked debate in the media about whether it can be relied upon to reduce or remove bias. Through Dweet’s system, brands can use an AI job description generator to post jobs to the platform. The team reviews the job postings and sets up a call with the brand to understand the nuances of the requirements. Talent across the platform can apply, and the Dweet team also recommends specific talent in their database and reaches out to people in the community to ask if they’d like to refer anyone. Applicants are vetted by Dweet’s proprietary algorithm — which compares the skills listed in each application to those the brand requires based on discussions between the Dweet team to decipher how good a fit candidates are for a particular role. The Dweet team speaks to applicants to clear up any missing information. Dweet recommends the top candidates with a summary of their application, and from there, the brand can begin the interview process. “It’s removing the biases with Dweet being the middle person, bringing fresh eyes and objectivity, which then allows brands to look at candidates with a renewed energy and vigour because they’re [recommended] through someone else,” says Daniel Peters, founder of consultancy Fashion Minority Report, which recently partnered with Dweet on a mentoring project. But some people have concerns that AI could repeat the biases of its creators. “AI models often learn from past hiring decisions and performance evaluations. It assumes, therefore, that those decisions were fair, equitable and free from bias, discrimination and in some cases racism,” says Shereen Daniels, HR strategist and managing director of racial equity advisory firm HR Rewired. Daniels says there are a variety of factors that should be considered: “What data set was used to build the algorithms? What controls are put in place to ensure that biased hiring decisions are not replicated at scale? What is the decision-making process that’s used to determine what is an acceptable and unacceptable candidate? Who is accountable for ensuring these factors are considered? What does the ongoing evaluation look like? Can AI providers evidence how they have addressed this and how they monitor to interrupt any ‘accidental’ perpetuation of bias and discrimination?” Piffaut says that, to his knowledge, there has not been a case where Dweet’s talent team has had to intervene in the automated matching process to override potential biases. Dweet discounts profile pictures, educational background, social media presence, and individual identity (such as gender or ethnicity) in the matching process, focusing purely on hard data such as skills, experience, previous job titles, location and availability. For Dweet, the tech side is just the first step in the recruitment process, and the human element will always remain. “Our automated matching is providing a much larger shortlist for our talent success team to filter down to a much more curated shortlist before the client receives talent profiles to review,” says Piffaut. Unsuccessful candidates are always contacted to let them know they have not been included in the shortlist. Dweet plans to further reduce the risk of bias from hiring decisions by offering sourcing tools and talent assessments to highlight why a candidate’s skills could be relevant regardless of their background; offering self-development and upskilling modules for talent; and completely anonymising profiles (currently, applicants can choose to anonymise their profile picture and secondary information such as education), allowing brands to only view relevant skills and experience until they’ve confirmed an interview. Top talent is increasingly challenging to attract and retain: according to a report by global HR platform Remote, employee turnover rates increased 8.7 per cent from 2019 to 2023, up 9.6 per cent in the US and 7.7 per cent in the UK. As such, experts agree a skills-based approach is preferable. “Employers should be valuing skills over qualifications and prior experience if they want to increase the pool of talent in front of them,” says Kate Holt, partner at KPMG UK’s people consulting division. “Many employers are using a range of assessment and gamification tools to attract candidates based on their ability to demonstrate a particular skill or solve an industry-specific problem over purely valuing a degree qualification.” By opening up the talent pool, Dweet aims to encourage brands to interview candidates with different backgrounds that they may not have considered. “We can’t remove the bias on behalf of our clients when it comes to the interview and selection process, but I think we can improve it over time by showing more and more why you should give people a chance because they’re competent in a particular area,” says Duane. The internal team has done some unconscious bias training with external consultants, but Duane says he hopes the company can accelerate internal diversity and inclusion training as Dweet scales. Scaling the tech, and the business Dweet’s long-term vision is for the platform to become more automated over time, which will prove more cost-effective for the startup and also its clients. According to Duane, the split right now is about 70 per cent tech and 30 per cent human (primarily for building relationships with clients and talent). The end goal is for Dweet to become a digital product that allows talent and hiring managers to connect, with the matching happening behind the scenes via the algorithms. The team will continue to monitor the outcomes but will become “more like relationship managers or career advisors, rather than the level of intervention they have in the recruitment workflow today”, says Piffaut. “We started our service as a managed marketplace and not a fully automated marketplace because our industry requires a human touch and some finesse.” “In a world that is increasingly moving toward automation and the inanimate, I think it’s about having something that is so human and at the same time so advanced in its approach,” says Lanvin’s deputy general manager Siddhartha Shukla, who has used Dweet to recruit a range of temporary and permanent positions. Piffaut says recruiting some roles in fashion, such as design, requires more of a human touch than others. Since expanding past retail, the Dweet team has had to deepen its tech capabilities and community reach – both of which are still ongoing developments. “Building our own matching algorithms, specific to the fashion industry, has been way more complex as we have had to train them on so many different job titles and subtlety,” says Piffaut. “At the same time, to be truly on-demand and to have enough talent available to meet the varied needs, we had to reach a certain community size.” Its focus on the fashion industry is a draw for clients. “What I like about Dweet is that they really understand the industry, which sometimes a [traditional] recruitment agency doesn’t,” says Olivia Sayag, who joined Valentino as HR director of North and Western Europe in 2022 after working at Lanvin from 2019, where she first connected with the Dweet founders. Dweet initially built its reputation by focusing primarily on senior-level roles but is now looking to offer more opportunities at the entry-level. Its recent project to match its brand experts with mentees from Fashion Minority Report’s community of emerging young talent was a precursor to Dweet’s long-term plan to develop Dweet Academy, a non-profit part of the business, to offer skills coaching to its talent community. Piffaut says matching senior talent with opportunities is easier than the entry level because it’s simpler to evaluate relevant skills, experience, past job titles and companies. Dweet is building and developing tools that will allow it to find these talents from outside of the industry, for example, from hospitality or airlines or straight out of school. “As our reputation grows and as the respect from clients grows, we start touching on more entry-level roles, which will ultimately allow us to help more young talent to enter the industry via Dweet.” Key takeaway: Dweet is using algorithms to automate some elements of recruitment, streamlining the luxury fashion industry’s recruitment processes and helping companies overcome a period of high turnover. As it scales its use of tech, retaining a sense of human touch will be important — as will continually reviewing the risk of bias in its processes and finding new ways to reach entry-level candidates. Comments, questions or feedback? Email us at feedback@voguebusiness.com. More from this author: Ann Demeulemeester looks within for new creative director Off-White gains new CEO in New Guards Group shakeup Former Diesel CEO Eraldo Poletto resurfaces at footwear brand Santoni

Make better business decisions

Make better business decisions