Breaking AI bias: Predictive analytics platform aims to eliminate racism in marketing

Artificial Intelligence (AI) is often touted as the future of advertising and marketing, promising to revolutionize the way companies target and engage with customers. The numbers don’t lie, with reports projecting a colossal surge in global market revenue for AI in marketing, quadrupling from $27.4 billion in 2023 to $107.4 billion in 2028.

It’s no secret that many companies are already capitalizing on AI to spruce up their marketing strategies. Using AI-assisted dynamic decision-making, marketers can fine-tune their programmatic media buying and determine the winners of in-market tests. The AI can also predict and identify potentially profitable audiences for marketing purposes.

And that’s just the beginning. AI is now taking on the role of content creator. Marketers are finally loosening the reins and allowing AI-generated content for media placements, beyond just search advertising. To create custom images and copy for creative testing, companies are feeding AI with inputs from various sources, including past performance, marketing ideas, analyst insights and specialist knowledge.

However, as we transition into this new era of AI marketing, we must also acknowledge the potential pitfalls. AI tools are not infallible, and they can contain inherent biases and perpetuate racism.

Addressing bias and racism in AI

A case in point was the recent controversy surrounding the partnership between Levi’s and fashion studio Lalaland.ai. The AI-powered digital fashion studio creates hyper-realistic fashion models with the aim of increasing diversity in the industry. However, the announcement was met with fierce criticism, especially from models of color who already face a lack of representation. Many questioned the decision to create artificial models instead of hiring real models, calling it “lazy,” “problematic,” and most importantly, “racist.”

“The future of marketing is going to be about datasets and not gut reactions or hunches,” Larry Adams, founder and CEO of diversity-focused AI platform provider X_Stereotype, told VentureBeat. “In the past, we used human intelligence — of mainly white males — to create advertising. Today’s approach uses datasets pulled from various sources to create marketing campaigns that extend to many more media channels. And yet, those data sources are still based on past performance, created and evaluated through the lens of white males.

“Artificial intelligence,” he went on, “is only as good as its inputs. Considering most of the data that exists is biased because of its one-sided point of view, the lack of diversity and gender experiences creates a gap in the knowledge base that AI draws upon to make its decisions, which impacts the output.”

With more and more reports coming out that AI can be racist, from marketing ploys like racist AI rappers to social media, false facial recognition in design to Facebook’s blatant ad placement discrimination, there’s a need in the industry for checks and balances — for a solution with human-informed insights from unbiased datasets.

That’s precisely what X_Stereotype aspires to offer — an AI-powered platform that analyzes content through a lens of diversity and inclusion, enabling marketers to identify racial bias and risk factors present in content at the earliest stage of development.

From tragedy to transformation

As a seasoned marketer, Adams was always invested in driving effective advertising through data, culture and insight. But it was the murder of George Floyd in 2020 that spurred his latest venture, X_Stereotype. As a Black professional in corporate America, Adams had long felt the challenges of navigating the industry while dealing with the unique experience of being Black in America. But it wasn’t until the Floyd murder that the marketing community at large started to take notice and ask questions, leading to a conversation around inclusion and authenticity.

Adams was also working as a senior advisor on Michael Bloomberg’s 2020 presidential campaign, where he was responsible for content strategy, marketing automation development and data strategy. He quickly realized that the campaign was not effectively reaching Black voters. Despite efforts to understand and reach multicultural audiences, stereotypes around how to connect with a Black audience were still prevalent, and the necessary data and messaging were not available.

To address this, Adams teamed up with entrepreneur Phil Alexander, who had been working on a research grant involving understanding how different human attributes impact working relationships. Together, they commissioned a full study with neuroscientists, psychologists and other specialists to quantify their findings and create algorithms that calculate 40 new psychometrics.

This research ultimately led to the creation of X_Stereotype, an engine designed to drive effective advertising while creating a less racist and more inclusive world, fueled by a truthful, inside perspective of today’s diverse experience.

Through X_Stereotype, Adams is taking his life’s work to the next level by combining data, culture and insight to drive more effective advertising that truly resonates with diverse audiences.

“The resolution to bias in AI systems is to create new datasets that represent a multitude of aspects of our diverse audience sets,” he said. “In the case of X_Stereotype, we have generated 4 million (and growing every day) new data signals that represent the complexity of today’s multicultural society. X_Stereotype structures our data to make it easy to integrate with other datasets to help AI make better decisions and create more representative outcomes.”

AI-powered predictive analytics

X_Stereotype is an AI-powered platform that uses predictive analytics to gauge the performance of content across different races and cultures. The platform offers 40 comprehensive metrics for content, including emotional impact, bias, inclusion and purchasing intent.

Using real focus groups and natural language processing, the platform converts human responses into emotion and sentiment scores, allowing its AI to learn from and improve upon itself.

By recognizing bias in content, X_Stereotype aims to predict the public’s reaction to an asset, generating an insights scorecard that identifies potential racial bias and risk factors.

A single focus group to refine its AI model creates roughly 300,000 data points. Collectively, X_Stereotype’s market research efforts generated over 4 million data points. X_Stereotype also collected a broad range of consumer demographic data such as age, gender, household income and marital status.

Companies such as FanDuel, Chipotle, J&J, Bob Evans and P&G have used X_Stereotype’s platform to gain the predictive power of AI as they pursue their business goals. The platform has been used in many settings, from early concepts to commercials featured in the Super Bowl. The platform claims to double ad effectiveness while reducing costs by providing clear, actionable insights earlier in the process.

One notable case is that of Bob Evans, where X_Stereotype’s insights were appreciated and provided a sense of comfort in knowing that all their concepts were being reviewed.

With over two years of experience working with AI providers and large language models (LLMs), the company plans to introduce more generative AI to the platform in the near future.

“X_Stereotype is on a mission to remove bias from marketing,” said Adams. “The driving force behind our mission is our own experiences, plus our love for marketing, and opening the door for more representation by creating a marketing culture that values our input and sees that it’s good for society and business results.”