The use of AI is increasing, especially in marketing. Data from HubSpot’s AI Trends in Marketing report shows that 74% of marketers use AI in their work. With so many marketers adopting AI, it is important to recognize and address its known limitations, especially the biases embedded within them.
As an inclusive marketing strategist and consultant, I’m trained to recognize bias when I encounter it. When I review materials and campaigns for clients, I assess whether bias, cultural insensitivity, or inappropriate messages have crept into the communications.
But most marketers I’ve encountered either don’t have these skills yet or are actively working on developing them. Many often do not take into account the biases created by AI, which, if released into the market, could prove detrimental to the brand’s marketing efforts and reputation.
To help you use AI more responsibly and effectively, I’ve created some pointers to help you eliminate bias. Let’s dive in.
The fundamental question to convert more customers
Before we dive into the actual prompts, it’s helpful to make sure you’re familiar with the fundamental question consumers (particularly those from underrepresented and underserved communities) are asking.
As consumers evaluate whether or not your brand is right for them, they are consciously or unconsciously trying to answer the question: Is this product for someone like me?
Every part of your customer journey serves as input to answer this question.
So if any aspect of your customer journey is biased, you are signaling to potential buyers that “this product is not for someone like you.” In most cases, this signal is not what the brand wants to send.
Now let’s focus on how you can use AI to tell your audience that “this is for you.”
Joyann Boyce is an inclusive AI expert and founder of Inclued AI, a tool that helps marketers communicate inclusively. She told me that it was helpful to imagine AI as a very well-behaved puppy.
She explains, “It’s like someone has already house trained the puppy. And they’re going to give you the puppy and you’re going to adapt it to your home.”
When customizing your well-trained AI puppy, you need to train it to ensure it communicates with your customers in a way that draws them closer to you rather than repelling them due to prejudice.
Listen to the full conversation with Joyann Boyce on this episode of the Inclusion & Marketing podcast.
How to use AI to detect and eliminate bias
While AI systems can inadvertently perpetuate biases present in their training data, they can also serve as powerful tools for detecting and mitigating biases in human decision-making processes.
By leveraging AI’s ability to analyze patterns and identify potential inconsistencies, you can create more objective assessment frameworks and uncover blind spots that might otherwise go unnoticed.
Provide the correct context.
First, provide your AI tool with clear context about its role and perspective. Be specific about which persona or expert viewpoint you want the AI to take, and define the specific lens through which you want it to analyze your content when providing feedback.
Right off the bat, I like to tell my AI employees that they are a highly skilled, integrative marketing strategist.
This works from a general perspective for overall content review and for a wider audience. However, if you want to be even more specific about what type of feedback you’re looking for, adjust the context for this expertise point.
Let’s say you want to find out if an ad is targeting consumers over 50. You can ask your AI companion to review the content through the lens of a marketer who has expertise in this consumer group.
As an example, I asked AI to review a website about anti-aging products. I asked it to act like a marketer who specializes in reaching consumers over 50. The AI could then use this basis to give me helpful feedback.
Here is the first part of the feedback on a language that was age-sensitive and lacked inclusivity:
Here are the AI’s recommendations on how to improve the copy to make it more inclusive:
The aim is to prevent the publication of content that is already biased. However, there will be times when you need to re-evaluate and improve something that is already on the market.
Here you’ll find a prompt to help you create new content and one to help you improve content you’ve already created.
Prompt for review
You are an inclusive marketing strategist who specializes in marketing to consumers over 50. Please read these headlines and let me know if I see anything that might mislead consumers interested in the product. I’m particularly interested in anything that would be offensive, culturally inappropriate or simply not inclusive.
Prompt to create
You are a copywriter who has specialized in marketing to consumers for over 50 years. Please consider 10 headlines for this skincare website that achieve our product goals while making our ideal customers, including people over 50, feel seen, supported and part of our brand.
Provide information about your consumer.
One of the challenges many marketing communications face when it comes to being inclusive is that the brand has not effectively defined its ideal customer.
For example, a brand might say that it targets “women ages 25 to 34 who want to advance their careers.” But while there are specifics related to age, gender, and even desires, there is still so much context missing that could influence the way the consumer receives the messages you create.
Therefore, when working with AI, avoid treating it as if it were aimed at a general market audience. Instead, give your AI companion additional details about who you want to communicate with. This will help ensure messages are better tailored to the audience you are targeting.
So instead of saying you’re creating landing page copy for “women ages 25 to 34 who want to advance their careers,” add details about who they are. This additional information will help your AI companion create less biased messages.
Some of the identity-based details about your ideal customer to include in your prompt might include:
- Racial and ethnic identities.
- Sexual orientation.
- Religion.
- Marital status (e.g. married, children).
- Economic status.
- Where they live, because regional differences can affect the choice of words.
- Including people with disabilities and neurodivergence.
Therefore, a prompt to write copy for a landing page might look like this:
Prompt to create
The target group of this landing page are women between the ages of 25 and 34 who want to advance their careers. This includes Black, Latina and Asian women. Some are married. Some have small children at home. All have at least a bachelor’s degree, live in the USA, Canada, Australia and Mexico and 30% of them are neurodivergent. Most of them do not yet own their own home. Please design a landing page for them that takes their identity and needs into account.
Prompt for review
Please identify anything in the copy of this landing page that would prevent people with (insert identity) from being seen, supported, and belonging.
Be direct about what type of bias you want to eliminate.
Your target customers have many identities and you want to make sure each person feels connected to your offering. That means creating inclusive campaigns that avoid bias, and AI tools can help you get it right. All you have to do is tell your tools what type of content you need not want to include in your messages.
When I conduct an inclusion audit for my clients’ brands, some of the things I look for include:
- Including language.
- Performance dynamics.
- Depiction.
- Stereotypes.
- Identity-based friction in the customer experience.
It is particularly important to be so specific. If your AI tool only focuses on one area, you may miss other problematic areas.
In this example, I asked AI to evaluate a blog post for cultural bias. Some problem areas were missed that I would have pointed out. When I asked it why it hadn’t addressed these issues, it replied as follows:
Depending on the type of communication you are composing using AI, be sure to include specific instructions about the elements you want to include or avoid.
Prompt to create
Please create some images that we can use for this social media ad that reflect our ideal customers. Provide an appropriate representation of the diverse identities of the consumers we serve, free from common stereotypes and cultural biases.
Prompt for review
In this ad, please highlight any areas that would be considered problematic from an inclusion perspective. Consider inclusive language (e.g. Are AAVE included?). Are there stereotypes associated with the images we highlight?
It’s time to get rid of the bias built into your AI.
When your AI-generated content authentically reflects your brand values and meaningfully appeals to diverse audiences, you build trust and connection with prospects who might otherwise feel overlooked or misunderstood.
This integrative approach not only helps you avoid alienating potential customers due to unconscious bias, but also demonstrates your commitment to serving all customers equally. This can significantly differentiate your brand from the competition in today’s market, where consumers increasingly expect companies to be socially conscious and representative.
Retraining your AI tools to detect bias can help you reach a larger audience. The sooner you switch, the faster you can grow.