May 17, 2024 –
This MarTech Chat builds on our recent discussion of Artificial Intelligence (AI) in marketing. Tommi Marsans, Associate Director Business Transformation, Marketing Decisioning Science for Verizon Business Group, and John McElhenney, Chief Marketing Officer at iterativ.ai, were the featured experts. Martechify’s Tina Schweiger moderated the conversation.
Setting the Stage
This session focused on the various tools and approaches to using artificial intelligence. Key insights include pragmatic AI applications in marketing and the importance of ethical AI use to avoid bias and other unintended consequences.
Tommi has more than 30 years of experience in marketing technology, which she has focused on blending technology with strategic goals. As the leader of AI Decisioning at Verizon Business Group, she has led the way in creating innovative solutions that adapt to customer needs across different channels. Tommi’s creative use of technology has earned her several prestigious awards. Outside of work, she enjoys Sci-Fi, especially Marvel and Star Wars, and has a large collection of movie memorabilia.
John is a marketing technology expert and writer and brings a wealth of experience, including leading the first launch of Dell.com, which quickly became the world’s biggest ecommerce business. He is Chief Marketing Officer of iterativ.ai, which funds various startups in the AI and entertainment spaces. Their initial project, ClickThis.ai, is a dedicated team focused on enabling AI for marketing and sales using currently available tools. Outside of work, John’s band, Buzzie, is gearing up to release their fifth album in March.
May
Perplexity AI
Perplexity is changing the nature of search for end users. Perplexity AI is different from ChatGPT and other conversational AI experiences because it’s designed to provide search engine research.
When you provide a question or query, it still conversationally provides you with an answer to your question. In addition, it gives you all the links to the content it used to generate that answer.
This documentation of sources won’t necessarily eliminate AI hallucinations, which are responses containing false or misleading information presented as fact. However, it’ll help you spot when the AI is filling in the gaps of available information. You will also be able to identify when the sourced content is from a page or website written as satire, humor or any purpose other than education and information.
The transparency of source material is something that ChatGPT, Google’s Gemini and other standalone generative AI options have not offered. They’re now scrambling to catch up.
AI copilots
AI is also used as a companion inside several enterprise marketing tools. These AI companions are often referred to as “buddies” or “copilots.”
Microsoft chose to officially name its copilot “Copilot,” and the overlap of their branding and the general term is a bit confusing. Microsoft Copilot is an AI companion that shows up in Microsoft products: the Windows operating system, the Edge web browser, the Bing search engine and Microsoft Teams.
Other enterprise sales and marketing tools are using copilots as well. Salesforce Einstein AI, for example, is a copilot that helps you get information about your own customer and sales data.
GitHub has an AI copilot, also named Copilot, that can help with coding and development work. Similarly, the copilot inside Adobe Firefly is there to help you craft design prompts and generate all manner of digital artwork within Adobe products.
Different types of AI are used in copilots:
ChatGPT marketplace
Features available to users of ChatGPT Plus have expanded. As a paying user, not only would you have access to the full range of data sources of ChatGPT-4, but you could also save conversational threads, create GPT agents and access GPTs written for specific purposes by other brands and OpenAI contributors.
Diving into the ChatGPT marketplace, here is an example of the “Logo Creator” developed by a member of the OpenAI community. It’s an interface for DALL-E, a text-to-image model, to help you create logos. This GPT agent asks you several questions to guide the logo design, rather than you needing to have a logo idea already in mind.
Here is an example of the output from Logo Creator when prompted to create a logo for Iterative.ai.
AI as a service for enterprise
Pragmatic AI can be integrated and used across multiple enterprise platforms. It becomes a service layer between the various sales, service and marketing tools and the enterprise data lake. The AI provides context and consistency based on the massive amounts of behavioral data in the data lake and your third-party data sources.
When integrated into an enterprise this way, the AI can take in all the available data and ask for the context necessary to get to a specific answer about a specific customer and their likely product or service needs.
With AI processing all that data for a specific customer, hyper-personalized website experiences become possible. AI, as an enterprise service, is capable of generating dynamic website experiences for customers. This can include personalized images, prioritized actions, the products or service offerings most likely to be purchased by that customer, and bespoke messaging that’s likely to resonate.
Generative AI can be integrated into the marketing workflow today. As it exists now, AI can be used to aid demand generation with site design, content development, search engine optimization and paid campaign optimization.
AI can also be used in email personalization, metric analysis and some UX design to improve onsite engagement. Inside your CRM, AI can help with segmentation, lead scoring and customer requests for more efficient lead nurturing. Put simply, AI can be integrated into almost every aspect of marketing operations.
The ethical use of AI is important to follow, particularly with regard to proprietary data security and transparency of data use. When using an open-source AI like ChatGPT to generate content based on data you’re putting into the system, you have to remember that your data input becomes available for the AI to use publicly.
Also, consider that if you restrict AI to only using your data and not outside sources, there is inherent bias in the output.
Two big risks with using AI:
To be ethical in the use of pragmatic AI, make sure you’re neither excluding nor privileging any particular group with the algorithm. Here’s a human resources example: If you train the AI based on who you’ve hired in the past, you run the risk of the AI filtering out job applicants who are not like existing employees.
Ethical AI use also requires that you “do no harm.” If you oversimplify the objectives, AI could make choices that impact quality and safety in order to lower production costs. If you simply tell it to sell more, the AI could lower the price until you have no profit margin because it’s optimizing for volume rather than profit. For that reason, exact prompts are critical—the more specific you can be, the better the AI output.
Get the most out of the insights Tommi and John shared by watching the video of the entire session, including the audience Q&A portion.
Visit Martechify’s Resources area to view more content on AI applications for marketing.
