July 24, 2023
Source: Energy & Management Powernews
Artificial intelligence can also take on more and more tasks in the energy industry. But its use is not without pitfalls.
What happens when you ask an artificial intelligence (AI) how the use of AI will change the energy industry? Here's how she answers: "Artificial intelligence (AI) has already begun to change the energy industry in many ways, and is expected to have an even more profound impact." At least, that's how the chatbot ChatGPT, developed by the U.S. company "OpenAI," puts it.
In fact, according to a study by the German Energy Agency (Dena), around three-quarters of all executives in the energy industry expect positive effects from the use of AI. They expect new, innovative business models, productivity increases and a more sustainable energy industry. However, the same study also states that only 17 percent of the executives surveyed felt well or very well informed about the topic of AI.
Moritz Brunemann, Management Consultant at IT service provider Arvato Systems, has been dealing with the possibilities of artificial intelligence since 2017. At the time, he was employed as head of private customer service at Stadtwerke Esslingen. "We had a large amount of customer data, and I wanted to find a way to evaluate which customers were at risk of termination." Together with a service provider, he developed a program to analyze the volumes of data.
It searches through all the customer data based on various characteristics, compares it with the data of customers who have already cancelled - and provides a list of those who might cancel soon. "I think that's the big value-add that self-learning programs provide: Recognizing things in an enormously large mass of data that humans alone wouldn't see," Brunemann said.
Rethinking required
The potential applications of AI in the energy industry, from customer interface to service and maintenance to grid control, are enormous. But they first require a fundamental rethink. This is because while conventional computer applications provide accurate solutions based on predefined rules - a calculator normally provides the correct answer "four" to the question of what two plus two is - AI models independently combine data using statistical and mathematical models. In this way, they calculate the result that has the highest probability of being correct - although it can also be wrong.
Especially when complex algorithms are used, it is almost impossible for the user to understand how the result was arrived at and why it might be wrong. In addition, the reliability of the result depends on the quality of the data with which the AI has been trained.
"Particularly in the area of forecasts," says Moritz Brunemann, "I have not yet seen a ready-made solution that is suitable for everyone. Regionality also plays a role here: an AI model from Munich does not necessarily work in Hamburg, and a model developed for an area supplier does not deliver optimal results in a large city. You always have to develop that very regionally or even individually for the energy supplier."
In addition, the data must also be updated again and again. Take climate change, for example: a weather forecast model developed on the basis of weather data five years ago no longer necessarily delivers correct results even today.
"The full extent of possible sources of error brought about by digitization and the use of AI cannot yet be clearly estimated," the expert report "Data Analyses and AI in the Electricity Distribution Grid" prepared by the Fraunhofer Institute for Energy Economics and Energy System Technology 2022 states: "Since AI applications (...) can usually use larger amounts of data and potentially new data sources that are not used by classical methods, AI methods may have a higher degree of susceptibility to data errors." However, AI methods themselves also offer the possibility of checking data for validity and consistency in advance of use,
AI will simply move into everyday life step by step
Already, many utilities are experimenting with the potential applications - to varying degrees. Stadtwerke Hamm (North Rhine-Westphalia), for example, has been using artificial intelligence in its customer center since the beginning of the year, as a "small support for telephony," as it puts it. Essen-based energy company Eon, on the other hand, is already using AI on a large scale - for the optimal alignment of wind turbines, for power grid control or for predictive maintenance of medium-voltage cables.
But in principle, says IT expert Brunemann, he does not assume that all utilities will now have to deal with all the technical possibilities themselves, increase staff, build up expertise and make the corresponding investments. "I believe that AI will simply move into everyday life with us. That energy suppliers will also find AI modules in their software and then use them - or not."
So, in response to an E&M inquiry, software provider SAP reports that the SAP S/4HANA solution, which is widely used in the industry, already offers more than 50 AI-supported business scenarios. Starting with the automation of mass processes, such as the plausibility check of meter reading results or the identification of incorrect billing, to the improvement of plant availability and the reduction of maintenance costs through predictive maintenance, many applications have already been integrated, he said.
Employee support through AI-based clearing suggestions for reconciling bills and payments is already possible, as is the use of intelligent algorithms that can analyze and answer customer inquiries on all channels in the area of customer service or identify and inform the responsible employee.
The big task for energy companies now, according to IT expert Brunemann, is to determine how to deal with the possibilities: Should they be used? If so, to what extent? And how do you evaluate the results generated? From data protection to IT security, a whole series of questions are open here that have not yet been clarified at the political level either.
And ultimately, there is also the very big question of responsibility: It would already be technically possible, for example, to leave power grid control completely to an AI. But, says Brunemann, "If the AI then makes mistakes and the entire grid has to be shut down, then we'll have damages in the millions. And I can't very well sue an AI."
And so ChatGPT also concludes its response to the question about the potential uses of AI in the energy industry by saying, "It is important to note that these changes do not come without challenges. Privacy, cybersecurity, and ethical considerations are important considerations when using AI in the energy industry. Nevertheless, AI has the potential to significantly improve the efficiency, reliability, and sustainability of the energy industry."
Author: Katia Meyer-Tien
Accessibility
Language
