I have been blessed with a 25+ year career that is varied and rich in experiences. It has been a career (mostly as a consultant) that spans a diverse set of industries inclusive of manufacturing, defense, insurance, financial, healthcare and now utilities.
While the industries I’ve worked in have been quite different, there have been some similarities. One of those similarities I was recently reminded of, is that organizations often misunderstand the possibilities of positive disruption that new technologies and techniques can offer.
For example, when I worked on computer systems for a picture film processing manufacturer in the late 1990s, a common refrain I heard from the senior staff was, “customers will never want to take pictures with a digital camera, they will always want to have physical tangible pictures they can hold.” ………… Not too many people getting film developed anymore……….
A second example, in the mid-2000s, I worked for an organization that only used Lotus software for spreadsheets, word processing, etc. I remember hearing staff say, “Microsoft Office will never be adopted, Lotus is far superior, it’s been around longer and it is easier to use.” ………. How many spreadsheets have you put together in Lotus 1-2-3 lately? Some of my colleagues younger than me aren’t even familiar with Lotus!
As an evangelist for Data Science, Artificial Intelligence and Machine Learning I was reminded of the above recently at a couple of recent industry events where I overheard a comment along the lines of “Utilities don’t want and don’t care about Artificial Intelligence (AI). They never will.”
When I heard this comment, I was saddened! I wanted to jump in and say that the statement was simply untrue, however I understood their opinion was that of misunderstanding and requires education and grace. Additionally, I pondered how much future monetary savings organizations will miss out on if they adhere to that opinion over the next few years.
First and foremost, people don’t buy or use products just because they have AI in them. They buy and use products because they solve a problem.
For instance, Tesla owners do not buy cars with automated driving and parking features because they use AI. They buy the cars for the automated driving and parking features, that just so happen to use AI.
Using improved techniques offered by AI, I feel that utilities would love to have accurate automated processes in which AMI smart meter data could be used to:
- Identify which household appliances are turned on/off within a house connected to a smart meter.
- Identify when a pipe connected to a smart meter is very slowly getting “choked” by particulates over time (like limestone build-up).
- Identify which meters amongst the thousands each utility manages have had an anomaly that is out of the normal relative to its historical patterns (i.e., being smarter than using conventional threshold approaches).
- Forecast the effects of upcoming weather events on their smart meter networks.
- Identify when data in computer systems are mislabeled by staff (such as, an incorrect meter size or type of meter).
- Design a system that self-learns good behaviors versus bad behaviors to help with tuning and the loss of organic organizational knowledge through the impending retirement of baby boomers from the workforce.
Now you may be thinking to yourself, “Wow that sounds awesome! If someone can do the above to save my staff, customers, and the utility overall time and money, let’s do it! What can I do as the next step?” Here is what I propose:
- Look at the computing processes your customers and/or staff must execute today. Find the ones that:
- Involve a lot of manual analysis to find the “needle in the proverbial haystack.” You know the ones where your staff roll their eyes and tell you how many computer systems, spreadsheets, departments, etc., they had to go through to get to the answer and how much time it took. They would gladly give these tasks to someone else to do if presented with the opportunity.
- Entail looking for trends that are not easily seen by the human eye.
- Require a lot of hard-to-replicate-tuning knowledge from staff members that have many years of experience, are close to retirement and can’t be replicated by others easily. You know, the ones that keep you awake at night thinking about what happens if that employee leaves.
- Press your staff, vendors, partners, and yourself on what could be done to solve these problems. Then ask if there might be any AI techniques that they know of which could be used to help more fully automate the challenges above. If they answer “no” to those questions:
- Probe into how they know that AI techniques may not be helpful.
- Assuage fears that Hollywood has ingrained in us about how AI will take people’s jobs or do dastardly bad things. AI in the space of utility management is just using advanced mathematical/statistical techniques to help further automate mundane tasks that humans would prefer not to engage in.
- Set a vision for automation that involves how their work lives could be improved through automating the mundane and frustrating.
- Communicate using metrics like amounts of water, revenue, labor, truck rolls, time, etc., saved.
- Be on the lookout for dismissive comments against new technologies like AI, don’t accept them.
- Lastly, if needed, reach out for help! Sensus has a Data Science department that is aligned with product management, both of which are willing to have a conversation around AI.
We hope this blog has been helpful to inspire a different way of thinking about the relationship of AI to utilities. Oh, do me a favor! If you see anyone out there still using Lotus 1-2-3 spreadsheet software, take a digital picture of it and send it to me at firstname.lastname@example.org! 😊