Guarding Against the Tempting Allure of AI: Ensuring Bias-Free Progress

Guarding Against the Tempting Allure of AI: Ensuring Bias-Free Progress

Efforts are essential to address bias in data powering AI tools to prevent regression in the industry's advancements.


Source: Shutterstock

The use of AI in marketing is rapidly growing, with applications ranging from predicting personalized content to optimizing advertising campaigns. Brands are even training their own large language models to align with their unique brand identities. The global market size of AI in marketing is expected to reach $72.1 billion by 2030, a significant increase from 2022.

However, with every boom comes risks. For instance, Under Armour faced backlash earlier this year for an ad featuring British boxer Anthony Joshua. The ad used recycled footage from previous commercials and was criticized for its black and white, fast-cutting montage.

Critics were quick to question the ethics of reusing old work. Many brands also have concerns about accidentally using content that violates copyright laws, or about sharing customer data with AI systems that could potentially benefit their competitors.

These risks have led marketers to include clauses in agency contracts that restrict the use of AI without permission. To address these challenges, the Advertising Association took a proactive step by creating an AI task force last autumn. This task force is dedicated to assisting the UK industry in understanding and navigating the opportunities and risks associated with AI.

I wonder if enough attention is being given to how marketers can effectively use AI while also addressing gender bias and distortions within AI. This bias can affect AI-generated insights and recommendations for marketers. If we don't act quickly and diligently to eliminate bias in AI used in marketing, we risk undoing the progress made in how women are targeted in advertising.

The issue is particularly evident in the creation of visual content. Many AI models for generating images are trained on datasets from the internet that often portray unrealistic and stereotypical images of women. Rhonda Hadi, an associate professor of marketing at Saïd Business School, University of Oxford, highlighted that these datasets frequently showcase young, conventionally attractive women in sexualized or subordinate roles. When marketers utilize these AI tools for visual content creation, they may unintentionally reinforce harmful gender stereotypes. (Kudos to Dove for pledging not to use AI-generated imagery depicting women in their advertising and communications).

AI can exacerbate existing gender bias, but it can also play a role in filling important data gaps caused by gender discrimination. According to AI specialist and former Microsoft CMO, Candina Weston, addressing data bias is crucial when using AI on a large scale. It is a continuous process that requires ensuring that the data of the target audience is at the right level to produce the desired outcomes, whether AI is involved or not.

It is a challenge to rely on data that is meant to represent women today, as it is often incomplete or oversimplified. This can lead to AI being used in ways that are not accurate for our industry. One possible solution to this lack of data is the use of synthetic data, which is artificially created by algorithms or simulations to train machine learning models.

AI can both reinforce existing gender biases and help fill important data gaps caused by discrimination. This shows that the debate around AI in marketing is not clear-cut. Tamara Rogers, global CMO at Haleon, believes that marketing's use of AI should balance ambition with caution.

While her teams have utilized AI tools like CreativeX to enhance marketing evaluation and analysis on a large scale, across numerous creative assets, she emphasizes the importance of focusing marketing and advertising investments on the most effective areas. This rich level of insight ensures engagement and resonance with the target audience, aligning with Haleon’s goal of promoting greater health inclusion globally. However, there are limitations to where AI can be effectively utilized in the campaign journey. For instance, AI struggles to replicate the originality and creativity that human minds bring to content creation, especially in capturing local cultural nuances within campaigns.

The ultimate safeguard that marketers and their agencies must prioritize and strengthen is the last-mile audit process until data sets are comprehensive and free from biases.

So, it seems that generative AI is not as big of a threat as some creative agencies may have thought. Instead of worrying about this existential threat, maybe our industry should focus more on addressing the bias in the data that currently fuels many AI tools used in marketing. By fixing this issue, we can move closer to achieving the ultimate goal of our industry: creating personalized marketing for individuals on a global scale.

According to Weston, improving the quality of data is crucial for moving towards hyper-personalization. This aspect is often overlooked but is essential for filtering out irrelevant content and delivering more helpful information to consumers.

Let's be real, as a working mother who has been historically overlooked, I wouldn't mind receiving some hyper-personalized and helpful marketing.

As we work towards more inclusive and unbiased data sets, marketers and their agencies need to prioritize the last-mile audit process. This involves a human review to ensure that any content sent directly to customers is free from bias (assuming the humans involved are also unbiased). We should also be mindful, like Dove, of where and how we use AI in marketing campaigns. We must not be tempted by the allure of AI and risk undoing the progress we've made in removing bias and stereotypes from our work.

Editor's P/S:

As AI becomes increasingly prevalent in marketing, it's crucial to address the ethical concerns and biases that arise with its use. The article highlights the risks associated with AI, such as the potential for copyright infringement and the reinforcement of harmful gender stereotypes. Marketers must be vigilant in ensuring that AI is used responsibly and ethically, and that data sets are diverse and free from bias.

The article also emphasizes the importance of human oversight in the use of AI. While AI can provide valuable insights and automate tasks, it's essential for marketers to retain control over the creative process and ensure that AI-generated content is aligned with their brand values and ethical standards. By striking a balance between ambition and caution, marketers can harness the power of AI while mitigating the risks and promoting inclusivity and fairness in their marketing efforts.