• Home
  • News
  • Fashion
  • Female-led start-up removes the bias from artificial technology in fashion

Female-led start-up removes the bias from artificial technology in fashion

By Jackie Mallon


Scroll down to read more

Fashion |Interview


Artificial Intelligence has been plagued by accusations of inaccuracy and of being a vehicle for racial profiling. Perpetuating the biases of its programmers, who are often young, white Silicon Valley males, the technology seemed to have become just another gatekeeper to prevent the already marginalized from being considered for jobs or advancing in the workplace. The algorithms are said to have significantly impacted racial and gender diversity in particular.

But one female-founded Toronto start-up operating in the fashion and beauty sector is betting on AI to help shape a fairer future. Halt AI co-founders, Amanda Cosco and Roya Sedigh who have extensive backgrounds in technology and fashion found they were regularly auditing algorithms for bias. And so the seed of Halt AI was sewn. FashionUnited speaks to Cosco about Halt AI’s latest launch, the Diversity Analytics Dashboard, which provides detailed analytics on the diversity of over 500 fashion brands, influencers, and retailers.


“Most of the stories you hear about AI are cyborg doom and gloom and kind of scary especially when it comes to facial recognition,” says Cosco. “But if we can use AI for good in order to improve diversity then it’s a pushback against that doom and gloom narrative.”

They identified that most brands measured their diversity from a hiring perspective but were neglecting their marketing and original content so they decided to use their technology to offer brands a snapshot of what their social media and marketing looks like from the outside. This stoked brands’ interest. “Brands were asking for contextual information, like how their efforts compared to other brands,” explains Cosco. “If they scored 7 out of 10 on race, well is that good?”

This motivated the creation of a benchmark for companies which evaluated their performance not just alongside their peers but within the industry in general. Among other criteria, it measures six diversity dimensions – race, age, gender, body-type, skin tone, and disability - for each brand across its Visual Content.

Pre-funding, and having received a number of grants from the Canadian government along with angel investment, Halt AI’s clients currently include Fortune 500 companies. The founders began with fashion and beauty because both are industries with high impact particularly on women. Says Cosco, “The technology is ready to scale into other industries, it’s just about building the brand reputation right now.”

A handful of content auditing services already exist, such as the Geena Davis Institute which tracks gender representation in media and entertainment. But Cosco says they are expensive, not scalable and riddled with human error. “What makes Halt unique is that we have an accuracy rate of over 90 percent and we can be quick, processing entire industries within weeks.”

Luxury fashion still lacks racial and disability representation

Only a year old, Halt AI data quickly confirmed that in luxury fashion, for instance, the industry is still very white, very young, with little representation with regards to disability, and while there is diversity in terms of race, the skin tone tends to remain light. “Post BLM and the George Floyd murder, companies made a lot of promises and there has been a lot of attention on that community, but the disability community is next to be spotlit,” says Cosco. “Athletic brands like Nike perform very well in terms of diversity whereas it’s the usual suspects, like Chanel, that don’t have a lot of diversity.”

But Halt Ai is not about finger wagging. The quantitive analysis they carry out gives companies raw data so they can make decisions within their marketing. Right now Cosco believes there is an absence of insight which leaves brands with a huge blind spot.

With NDAs in place Cosco is limited in what she can say about specific clients but there are some general patterns she has observed. “Brands are generally surprised by their lack of representation, particularly within disability, it’s probably something they just don’t think about,” she says. “We’re trying to build for them analysis so that they can track improvements over time to understand better, but helping them understand where they sit within the larger industry, telling them they are within the tenth percentile or ninetieth percentile has proved useful."

A pilot program for a sub brand within a large beauty company proved so successful that it is now being rolled out internally across all its brands so that all marketing teams can access it in their decision making. While the technology is a B2B tool at the moment, Cosco believes it will become a B2B2C because the consumers are the ones who will experience the benefits.

“We want to become the global standard for measuring diversity across every industry so that we can provide a seal of approval, a threshold of diversity that is acceptable for advertising,” says Cosco. “Right now there’s really nothing like that.”

Artificial Intelligence
Halt AI
Inclusive fashion