Harnessing the Power of AI for Regulated Utilities

Artificial Intelligence (AI) is hot right now; you’ve probably seen AI-generated art, heard an AI-generated song, or read an essay written by AI. But did you know that it can also be applied to the regulated energy industry? Understanding how to apply the benefits of AI to regulated energy companies is vital for the energy transition and society’s future. That is why HData is leveraging the power of AI to digitize and automate the regulated energy industry.

Simply put, AI is a technology that allows computers and machines to think and act in a way that mimics people. In the first half of the 20th century, British logician and computer pioneer Alan Turing started developing and experimenting with smart machines. These machines have been iterated on and form the basis of computers and much of the AI we know today.

Whether it’s the voice assistant in your phone or the help desk for your bank, society has been using AI to generate great answers  and results. Generative AI, like ChatGPT and DALL-E, is changing the game even more by going a step further. These systems can now be creative. 

The next step in understanding AI is understanding how to control creativity when the machine diverges from an exact science. Artificial Intelligence is continuing to evolve, and as it does, we at HData are focused on harnessing its capability to ensure energy companies can be more efficient in understanding, analyzing, and deriving insights from the tens of thousands of FERC reports that are accessible through our platform. 

HData created our own AI bot to assist you as you examine thousands and thousands of FERC data points. This tool implements generative AI into our existing solutions, allowing users to submit a simple query and receive an answer in seconds. Within HData’s proprietary product suite, we have multiple technologies we leverage to deliver all the features you love. Some are like ChatGPT – providing a conversational like interaction between two people or a dialogue.

Others are being built to solve specific pain-points like finding synonyms to words, generating a single answer to a question or even alerting users when a certain topic is mentioned in a document. Through this journey, we’ve landed on some really interesting capabilities, specifically, our new AI offering: The Librarian. It is designed to work within the HData Library to analyze FERC reports using generative AI capabilities. With The Librarian, you can easily access and analyze any Form 1, Form 1-3Q, and Form 714 contained within the Library with greater speed and accuracy. The Librarian leverages state-of-the-art AI technologies for more thorough data analysis, providing deeper insights and improving decision-making capabilities. Users converse with the Librarian and can ask follow-up questions based on the answers provided. 

So, what are the limitations? Are there any? Indeed, Generative AI and more specifically GPT has its limitations. Despite being one of the most advanced models available, it is still imperfect in understanding and interpreting human language. One of the main limitations is that it heavily relies on the quality and quantity of data it is trained on.

Thus, if the training data is biased, incomplete, or faulty, it could lead to inaccuracies in its responses. Additionally, systems like ChatGPT may struggle with context-specific knowledge or tasks requiring more complex reasoning, such as advanced problem-solving or critical thinking. That is why HData continues to test different ways of transforming how we store the FERC reports in our data stores while maintaining their actual context and content. We are also trying techniques such as Semantic Search and different Large Language Models to optimize how AI understands our domain-specific documents. 

Some might question the reliability or ability of our system to accurately analyze, understand and provide answers based on the FERC reports it analyzes. That’s why it is important to verify. You can trust the Librarian but should always confirm its results. The Librarian links the relevant documents in its answers for ease of verification. Just like all AI, the Librarian is still in training mode and it is always learning. We employ different techniques that help the Librarian decide the best result and ultimately present that to the user. We also built a feedback mechanism into the system, allowing users to provide positive and negative reinforcement to the Librarian. Such reinforcement is later used to fine tune the system. 

We know that AI is not a silver bullet and the journey is far from over. We must be prudent when building ChatGPT-like capabilities for users to better understand FERC reports. After all, many of our users are the key individuals carefully stewarding this data from each organization to the FERC. Data is meant to be used, and we are continuously working to make it even more accessible.  

This breakthrough technology in the regulated energy segment is revolutionizing the speed and accuracy with which companies can analyze data and make informed decisions. For example, our new AI platform is able to easily answer industry queries like: 

• "Give a short summary of any riders added to rates from the companies in the data."
• "Does this document mention legal issues that Company X might be experiencing?”
• "Give a short description of the companies that entered information on infrastructure replacement programs and the amount of money under those programs."
• "Are there any complaints filed against Company X  or its parent organization, Company Y?"
• A follow-up question to the above one is, "What did the Department of Environmental Management do?"

As our industry navigates the wild west of Artificial Intelligence, HData is on the forefront helping industry folks capitalize on the many opportunities it provides. See for yourself and explore our AI offering here.

EXAMPLE:

18/04/2023, 08:40:28,"What services do you provide?"

18/04/2023, 08:40:33," As The Librarian, I provide a range of services, including answering questions related to the HData Library and its offerings."

18/04/2023, 08:40:48,"What is your name?"

18/04/2023, 08:40:51," My name is The Librarian."

18/04/2023, 08:41:11,"What is plant Daniel?"

18/04/2023, 08:41:15," The HData Library provides information about Plant Daniel, including its 50% undivided ownership interests, it's potential restructuring, its potential retirement date, and the potential impacts of operating the units on an individual basis."

18/04/2023, 08:41:40,"Can you please share this information with me?"

18/04/2023, 08:41:48," Yes, Plant Daniel is a coal generating unit that is jointly owned by Mississippi Power and Gulf Power. In conjunction with Southern Company's sale of Gulf Power, Mississippi Power and Gulf Power agreed to seek a restructuring of their 50% undivided ownership interests in Plant Daniel such that each of them would, after the restructuring, own 100% of a generating unit. On April 24, 2020, Mississippi Power and Gulf Power entered into an agreement to transfer ownership of Plant Daniel to Mississippi Power. Mississippi Power has the option to purchase Gulf Power's ownership interest for $1 on January 15, 2024, provided that Mississippi Power exercises the option no later than 120 days prior to that date. Mississippi Power is assessing the potential operational and economic effects of Gulf Power's notice. Mississippi Power and Gulf Power are continuing negotiations on a mutually acceptable revised operating agreement. The impacts of operating the units on an individual basis continue to be evaluated by Mississippi Power and any transfer of ownership would be subject to approval by the FERC and the Mississippi PSC. The ultimate outcome of this matter cannot be determined at this time."

Curious to see first-hand what HData products can do? Request a copy of our ebook to learn more.