How AI Can Solve The "Silver Tsunami" And Support Small Water Utilities

By Christian Bonawandt

While far from prolific, applications of artificial intelligence (AI) in the water and wastewater industry are nothing new. AI and machine learning have been used for data analytics for years. However, for small utilities and those with an aging workforce, these tools seem too high tech and costly to be practical. This doesn’t have to be the case, though. AI tools — particularly generative AI (Gen AI) and large language models (LLMs) — are able to address critical workforce shortages and resource constraints within the water and wastewater industry.

For many utilities, especially smaller ones, resource scarcity is a constant challenge, limiting their ability to adopt sophisticated digital technologies. As such, the evolution of LLMs is a game changer for those with limitations in both manpower and capital. Furthermore, the industry is grappling with the “silver tsunami” — the loss of expertise due to retiring operators. These tools offer a unique path forward for small utilities, potentially accelerating their digital transformation and in some cases, even leapfrogging more widely adopted solutions.

Democratizing Capabilities

The primary disruptive force of Gen AI is its accessibility via human language, which fundamentally democratizes the use of this technology. This means water utilities don’t need data scientists or software developers to use such tools. Instead, employees can interact with data and make requests using normal human speech and text.

Moreover, the financial barrier to entry is low. For example, the Microsoft Copilot subscription is only $30 a month per person. Utilities can adopt one of any number of LLMs by signing up for an enterprise version and utilizing the security levels applied to other digital applications, such as turning off data sharing, to protect sensitive information.

Addressing The Workforce Gap

Fear abounds that use of AI will supplant human labor. However, most AI tools function less as a job replacement than as a job assistant. In fact, nearly every employee, regardless of their role, could use these tools in some form or another to make their job more efficient.

Another key value of AI is in capturing institutional knowledge from experienced operators before they retire, essentially creating expert systems in the wake of the silver tsunami. For example, the Water Research Foundation (WRF) recently created a chatbot to capture the extensive skills of a long-retired operator. Titled What Would Jerry Do?: Chlorine, the project trained the AI on hours of interviews with Jerry Kemp, an 82-year-old operator with more than four decades of experience. Water utilities across rural Georgia often turned to Kemp for advice during his tenure and after retirement. In addition to Kemp’s expertise, WRF also provided the AI with wastewater-related regulations, water quality data, and other relevant technical information. The result is a tool that operators can easily pose questions to and get detailed answers that include a “pro tip from Jerry” in his own words.

Another example of an initiative to preserve and curate industry knowledge using AI is an internal training tool implemented by DC Water. This application organizes videos of experts, such as SCADA operators, describing their procedures and then generates quizzes to test for comprehension as well as create searchable content.

Leveraging Unstructured Data

Another benefit of LLM is its ability to process traditionally inaccessible unstructured data, including images, audio, video, PDF files, and handwritten notes. LLMs use optical character recognition (OCR) to convert handwritten notes and daily logs into digitized output. The AI can then structure the data into an Excel spreadsheet or similar format. From there, users can leverage the human language interface to request specific outcomes, such as data cleansing, correlation with other data sets, and more.

LLMs can also incorporate computer vision. Some utilities are experimenting with feeding camera footage in the field processes and having AI process visual information in response to prompts or alerts. When asked, the AI can tell operators if a valve is open or closed, or if there is rust or a leak on a particular asset.

The Future Of AI And Water

While LLMs accelerate digital transformation, foundational work remains essential. Data governance is one of the most critical things that utilities have to think about before implementing AI tools. This is especially true given the ability of LLMs to make previously inaccessible data processable, underscoring the need for clear rules on how data is managed, accessed, and used. Training is also crucial for effective adoption. Beyond their knowledge-capture efforts, DC Water, for example, provided its employees with 12 weeks of training on the use of AI tools.

Nevertheless, gaining a competitive edge and maximizing the value of AI requires significant investment. To secure funding for enterprise-grade LLMs, utilities must focus on two key areas: fostering a community of shared learning and securing buy-in from leadership. Finally, to help utilities prepare for the inevitable rise of AI, they must start getting familiar with these tools now. The rapid rate of change and the rapid rate of advancement means utilities cannot afford to sit back. Instead, success lies in taking a proactive approach that will enable them stay ahead of the curve as AI continues to evolve.

To learn more about this topic, check out this recent Water Online Live event, featuring DC Water’s Dr. Robert Bornhofen and consultant Gigi Karmous-Edwards as they share how utilities are putting AI to work today.