When we began our journey into exploring the fundamentals of AI, we knew that artificial intelligence was already revolutionizing the field of marketing and transforming the way businesses engaged with their audiences. Artificial intelligence continues to evolve and innovate the ways we can combine prose and data into dynamic storytelling. From the visual allure of images to the persuasive power of language, AI technologies like multimodal models (MMMs) and large language models (LLMs) are broadening the capabilities of AI and shaking up the marketing world. As marketers, we’re interested in ways brands and their agency partners can leverage both tech advances into impactful marketing campaigns. So, which AI model is the right fit for your marketing mix? Let’s find out.

But first, a quick history lesson:

The complexities of different AI models (and the ridiculous number of acronyms you must become familiar with) can feel a bit overwhelming. To ground you in some fundamentals before diving into LLMs and MMMs, think of this as an AI “family tree” to understand the progression and interconnectedness of different AI technologies over time.

Classical AI (1950s-1980s): Yep. AI has existed for longer than you think and continues in this classical format to this day, although its no longer the dominant approach in AI systems. These AI systems were designed to follow specific sets of rules, similar to how a computer program operates based on lines of code. Think of this iteration as one of those pesky “how-to” essays you had to write in grade school. It serves as a permanent set of instructions or a rule book that an AI would follow step by step without the need to learn any further.

Machine Learning (1980s-2000s): This era of AI ushered in supervised and unsupervised learning, where machines were fed data and learned to identify patterns and make predictions on their own without explicit programming. This paved the way for the development of neural networks (inspired by the interconnected structure of the human brain) which laid the foundation for many advanced AI models and techniques that followed. Think of machine learning as deep learning’s “older sibling.”

Deep Learning (2010s-Present): In this evolution of the AI family tree, deep learning built upon the foundation of neural networks and supervised and unsupervised learning developed in machine learning to pave the way for AI as we currently know it: generative AI, LLM, natural language processing (NLP) and MMM. These models are essentially “cousins” blending machine learning and deep learning tactics, with each branch and sub-branch building upon and expanding the capabilities of the previous ones. Generative AI focuses on creating new content, while NLP and LLMs concentrate on understanding and processing human language and MMMs specialize in processing and generating multiple types of data simultaneously. Today, deep learning and neural networks are everywhere in AI! They power voice assistants like Siri, help doctors diagnose diseases from medical images and even recommend songs or movies based on our preferences.

Moving on from the historical background, let’s delve deeper into LLMs and MMMs and explore why marketers should be aware of their capabilities.

Elevating Marketing Strategies With Large Language Models

If you were an early adopter of AI, you may already be familiar with the term “large language models.” A powerful application of NLP, LLMs are AI models trained on massive amounts of text data. Unlike traditional chatbots with pre-programmed responses, LLMs can generate different creative text formats, like poems, code, scripts, musical pieces, email, letters and more, as well as answer complex questions in an informative way. ChatGPT is perhaps the most popular example of this state-of-the-art language model, though there is an increasing amount of them in the marketplace. The major benefit of having this type of sophisticated AI model is that it has a wide range of applications across various industries, including marketing.

Some key opportunities LLMs bring to the marketing mix are their ability to eliminate writer’s block for copywriters by generating thought starters, automating customer service interactions or providing context to the content creation process for projects. These benefits and more can strengthen marketing strategies and allow brands to pivot in real time to meet the evolving needs and expectations of their target audience. Because LLMs can only operate within the confines of the large data sets they’re given, brands would benefit from partnering with agencies that can navigate the complexities of providing AI with great input to get useful outputs.

Brands wanting to dip a toe (or dive right into) the world of AI should be mindful that AI is still currently in uncharted waters regarding the world of unknown legal ramifications. Proceed with caution!

Understanding Multimodal Models

Multimodal models feel like the “Yes, And” to LLMs and NLP. They seamlessly blend various types of data, including images, text and audio, to generate richer and more dynamic content. Remember our AI family tree? Well, MMMs are the youngest cousin that gets all the hand-me-downs and learns the ropes from hanging out with its cool older cousins. MMMs are built to incorporate data from multiple AI tactics, generating more comprehensive and contextually accurate outputs. They can, for instance, analyze an image and produce a descriptive text, or understand spoken commands and generate a relevant visual response.

Delivery robots, like those used by companies such as Starship Technologies, use multimodal models to navigate through complex environments by combining visual and sensory data. Tesla’s self-driving cars are another prime example; they employ MMMs to interpret various types of data, like road conditions, traffic signs and pedestrian movements, to make real-time driving decisions.

In marketing, MMM opportunities are vast. They can be used to take social listening to a whole new level by allowing AI to analyze consumer sentiments by processing and interpreting multimedia content from social media platforms. This can open the door for brands to capitalize on valuable insights into consumer preferences and behaviors. Additionally, MMMs can streamline the content creation process by automatically generating captions, descriptions and even video summaries, saving time and effort for content creators. Brands looking to harness the power of MMMs should collaborate with agencies experienced in handling complex data sets to ensure optimal performance and results.

MMMs empower marketers to deliver dynamic content experiences that leave a lasting impression, but beware of potential legal issues.

TL; DR:

• AI has come such a long way! From rule follower to a powerful creative assistant, it’s now a great tool for marketers to streamline dynamic storytelling and savvy marketing strategies.

• Large language models (LLMs) like ChatGPT can assist in banishing writer’s block and efficiently enhance brands’ customer service so that audiences are getting the help they need, when they need it—just mind the legal minefield!

• Multimodal models (MMMs) are next-gen tools for creating richer, contextually accurate content, offering game-changing opportunities for brands in social listening and optimized content generation.

As the AI family tree continues to grow and evolve, so does the marketing landscape, and the opportunities for brands are boundless. Ready to harness the power of AI in your marketing strategy? Email Nate Tredinnick at ntredinnick@hiebing.com to set up a call.