is the time to teach AI your brand //

Andrew Frank, Gartner’s distinguished analyst and VP, offers a modest suggestion: Create a custom AI for your brand. Many marketing organizations are using solutions that include artificial Intelligence, while others are scrambling to find use cases for the readily available generative AI. What is the first application of it? Branding yourself.

Customization is an important feature. Frank says that marketing must move beyond “embedded” or “out-of-the box” solutions. Frank cites Gartner’s research, which shows that while 55% business leaders are considering AI for all use cases (the number rises to 71% when AI has been used for more than four year), marketing is ranked seventh on the list of top twelve business functions that leaders believe will benefit from AI.

Start with brand. Frank, presenting at the Gartner Marketing Symposium in San Francisco, made the case that a brand is a “fuzzy abstract” concept. He also pointed out the enormous progress AI has made, and specifically generative AI’s, to handle the fuzzy. Chat GPT is a good example of generative AI that sacrifices precision in favor of a more or less accurate and broadly relevant output. It’s easier for the AI to determine whether a story was happy or sad, than whether it was true.

Frank says that the ideal brand is not a single concept but a collection of images, colors, tones, moods and values.

Frank said, “You already have a brand. You care about it and have developed assets for that brand.” “That’s a great situation to start custom modeling.”

Frank does not want brands to start at the beginning. “That is out of scope for most organizations,” Frank said. ChatGPT, which is available from Google and Amazon, is one of many AI models that are foundational. It is best to use one of these AI models, then customize it with the brand’s data. Frank explained that it becomes a replica of the original model with your custom additions.

Human oversight and feedback should also be included, if only to ensure that brand values are represented.

The model does not need to be fed by humans. “The beauty of models like this is that you don’t have to understand concepts it extracts. It will do it for you. “All you have to feed it is a corpus with examples, and it will do all the subtle semantic links that we find it hard to think of.

Who is on the team? This project requires input from marketers, IT experts, data scientists and AI specialists. Frank calls the role of the Model Owner the core of the team. Model Owners will not be data or AI experts, but they will be able interact with experts to translate their operational challenges into the needs of marketers. Frank explained that the role is not technical at all. It’s more a supervisory position that owns and articulates the training process. The person does not have to understand how the process of training works.

The model’s operational framework envisages that generative AI creates paid media, social ads and content, websites, apps, videos, and chatbots within the parameters set by the brand.

Why should we care? This is the most ambitious use case for AI that has been described so far. Without human involvement, it’s easy to imagine how branding can go awry. Frank is the first to admit that. Once you start introducing data scientists and an IT team with time to spare, it’s easy to believe that this is an enterprise-wide project.

Frank is confident enough to predict that brands will begin custom-training AI in 2026.

The article Teach AI about your brand first appeared on MarTech.