This guest blog post was written by Luba Abrams, an independent marketing and change management consultant who worked at BGE for 13 years.

AI. Everyone is talking about it. And it will transform how utility marketers perform their jobs. Most utilities are already using some form of AI, likely generative AI (feeding a training set into a tool so it can “learn” to make true/false decisions about that data). The potential benefits for your marketing team are huge, but there are still many caveats to consider before using the technology.

As with so many evolving tools, there are challenges to be aware of when using AI to streamline, analyze, customize, and perfect your marketing efforts. If you plan to use AI on your team, here are five limitations we encourage you to consider.

Robotic tone

AI-generated content can sound robotic without human intervention. Review all generated content to make sure it’s relatable and engaging so you don’t lose that human connection. Relying too heavily on AI can make your content seem flat and disconnected, which will be off-putting for readers. AI can also misinterpret translations, so it’s best to have someone who speaks the language double-check the tool’s output.

Data security and privacy

Utilities widely acknowledge that there are AI-related data security and privacy risks in all areas—not just marketing. Be aware of using personal data that utilities (and other websites) shouldn’t have access to. Mitigate risk by deciding what consumer information to protect and how.

Cost and complexity

Some uses of AI can be expensive. And the tools can be difficult to integrate, especially for smaller utilities. Smaller utilities may not have the systems in place or be able to capture enough customer data to make it statistically sound for use in modeling.

Pace of development

Tools are changing quickly, and AI training data isn’t always up to date. For example, Salesforce is retiring Einstein Opportunity Insights to focus on Einstein Deal Insights. And ChatGPT only uses training data through September of 2021, so much of what it provides is already outdated.

Potential for errors

Verify the accuracy of the data you put into AI models. You’ll avoid spreading misinformation or sending the wrong messages to customers. If the data inputs are correct, the AI-generated results will be more effective. The data must stand for all aspects of your customers, including their behavior and journey.

With all of the hype around AI, it can be difficult to find helpful information and sources you can trust. We love HubSpot’s AI Marketing—The Complete Guide. We also found GoDaddy’s Prompting 101: Writing prompts for AI useful.

Contributing Authors

Executive Consultant

Luba Abrams is an independent marketing and change management consultant. Prior to her role as a consultant, Luba worked at BGE for...