How Marketers are Utilizing Artificial Intelligence and Accounting for Data Bias

by Santiago Solutions Group

The business landscape is evolving faster than ever through the use and implementation of artificial intelligence. Since early 2021, companies have been optimizing their processes using AI for customer service, marketing, sales, supply chain, human resources, recruitment, analytics, financial, product development, and so much more. Businesses are finding that there are so many practical and creative usages for AI and are noticing that those who do not embrace AI in their standard practices will be left behind in the very near future. However, the everyday implementation of AI in business practices demands a high level of data accuracy to prevent data bias.

Today, businesses are using artificial intelligence to make better and more well-informed decisions, predict outcomes, improve marketing performance, mitigate risks, automate tasks, and so much more. Marketers have also started leveraging AI to increase the accuracy of research and campaign outcomes. According to an article by Hubspot, marketers are currently using AI for the following marketing practices:

• Content personalization
• Data analytics and predictive modeling
• Content generation
• Media buying
• Customer service
• Email marketing campaigns
• Forecasting sales
• Search engine optimization (SEO)

Within marketing and advertising, creative procedures can sometimes lean on generative AI because it speeds up the processes and minimizes the need for creative talent. According to an article published by Forbes, almost 3 in 4 marketing professionals are using generative AI in their day-to-day work. The article also goes on to mention that it is most widely used for the creation of content such as email and social media copy, images, blog posts, and SEO.

However, generative AI uses in marketing have a documented history of being insensitive and/or stereotypical of diversity and diverse experiences. Our SSG ChatGPT bot explains, “AI systems themselves are not inherently racist, but they can inherit biases from the data they are trained on, and the algorithms used to build them. This is often referred to as ‘algorithmic bias’ or ‘bias in AI.’” For example, our Chat GPT bot describes that AI has been known in many instances to perpetuate biases, due to data sets that already have historical biases embedded in them. In addition, bias in AI is also perpetuated by the algorithm’s design, reinforcement of biased behaviors by AI programs, and the data collection process.

As an article from TIME put it, “one of the biggest hurdles in incorporating AI is ensuring that companies have the right data to fuel models.” The article goes on to mention that according to a statistic reported by the Mobile Marketing Association, almost 8 in 10 said that they have trouble validating their data and its quality. Data accuracy and validation has been a huge problem in this analytical day-in-age. Which is why Carlos Santiago, the CEO and founder of SSG and co-founder of AIMM (The Alliance for Multicultural and Inclusive Marketing), has spearheaded the industry’s efforts in data accuracy and validation, alongside other partners such as Truthset and Sequent Partners. Data accuracy is more important than ever now that we are experiencing a generative AI boom. According to Truthset’s Benchmarking Study fielded in 2022, accuracy for Hispanic, Black, and Asian US audiences have a data accuracy of 83%, 80%, and 57%, respectively. This means that businesses, and most likely, artificial intelligence, are using data that is not fully reliable or representative of Multicultural audiences to make business decisions, and in the case of AI, the biased data is used to reinforce machine learning.

During the AIMM conference held in early October of this year, keynote speaker Rex Briggs gave a talk titled, “The AI Conundrum, How To Be Seen When AI is Biased.” During his presentation, he discussed the pros and cons that AI will have on the consumer marketing industry. Briggs displayed examples of AI generated in-language voice ads with unique calls-to-action and he claims that ads developed using AI will have more than 2x the impact at less than half the cost. Additionally, he makes the case for proper training, governance, and accountability for AI models in order to ensure that AI does not have a bias towards people of color. Briggs gives the example that when you type “bad doctor” on the AI image generating app DALL-E, images of black and brown will be generated using this command. This means that due to faulty and biased data sets, DALL-E’s image generative AI has been trained to associate a “bad doctor” as being someone of color, which reinforces race/ethnicity stereotypes. There is a pressing need for AI regulation and implementation of unbiased data when training these models in order to ensure that people of all races, ethnicities, identities, and abilities are represented accurately and fairly.

In addition to potential bias in data collection and transparency, policy makers are starting to get concerned with consumer privacy and generative AI’s internet scraping practices. Chat GPT tends to store consumer account information, conversations with chat bots, location, IP addresses, and device information. Users are also becoming more aware that Chat GPT is also sharing this information with other entities such as business partners, legal system, chat bot trainers, etc (Does Chat GPT Have Privacy Issues, Kate Veale). Needless to say, consumer information isn’t kept private on the generative AI web.

As we head into a new era of business and marketing that has been revolutionized by artificial intelligence, we must ensure that data accuracy is improved so that marketers can continue to make more informed and accurate decisions using unbiased data and data collection processes. Moreover, with accurate consumer data that has been validated, marketers will be able to make unbiased decisions for predictive modeling, content generation, media buying, customer service, email marketing campaigns, forecasting sales, SEO, and most importantly, content personalization, which is the marketing/advertising of the future. As marketers implementing AI into our everyday practices, it’s important to ask ourselves whether the implementation of artificial intelligence and its known biases will create setbacks for progress made for diverse representation, diverse experiences, voices, and sensitivities into marketing processes.

Sources:

• AI Marketing – The Complete Guide, Rebecca Riserbato, Hubspot, October 19, 2023
• How Marketing Executives Are Thinking About Integrating AI Into Their Strategies, Simmone Shah, TIME, October 18, 2023
• Artificial Intelligence is Revolutionizing Marketing. Here’s What the Transformation Means for the Industry, Jessica Wong, Entrepreneur, March 6, 2023
• Top AI use cases in marketing to elevate your 2024 strategy, Annette Chacko, Sprout Social, October 19, 2023
• AI-powered marketing and sales reach new heights with generative AI, Richelle Deveau, Sonia Joseph Griffin, and Steve Reis, Mckinsey & Company, May 11, 2023
• Does Chat GPT Have Privacy Issues, Kate Veale, Make Use Of, September 21, 2023
• The AI Conundrum, How To Be Seen When AI is Biased, Rex Briggs, AIMM Member Forum, October 2023