AI will transform the energy industry in time, but close regulation will govern progress

Energy companies, like other suppliers of services to the public, are using AI to streamline their interactions with customers, to anticipate, guide and respond to customer queries, seeking improved outcomes, faster response times and lower call centre costs. For example, energy companies are using AI-powered chatbots to handle routine customer inquiries, such as billing questions, service requests, and outage reports. 

They are also using AI, data and non-AI software in operational ways. Offshore generators increasingly use sensors, modelling software and environmental data to target repair and maintenance on equipment as it’s needed, rather than to a routine timetable.

They are using AI and robotics to actually carry out some of this work, repairing and maintaining off-shore facilities. This reduces cost and complexity, but also has health and safety benefits because it means fewer people are sent to sea to operate in high risk conditions. For example, the industry is developing autonomous underwater robots that can perform precision maintenance on offshore installations. These robots are equipped with advanced AI and control systems, allowing them to handle complex tasks such as visual inspections, cleaning, and repairs in turbulent sea conditions.

There can be a blurred line here between actual AI and what some vendors call AI but which is actually just software or technology that existed before. This can be due to the broad and sometimes ambiguous definitions of AI, which can encompass anything from simple automation to advanced machine learning algorithms. But from energy generators’ point of view the distinction is not that important as long as the right functions are being carried out safely and efficiently.

Learning from the wider infrastructure industry, energy companies can build digital twins of their major generation and transmission assets to help with improvements, repair, maintenance and upgrades. This can help to optimise operations, improve efficiency and reduce downtime.

Better AI-enabled weather prediction can help with planning to preserve equipment, and also can help predict patterns of generation of renewable energy to help with grid and distribution decisions, ensuring a stable and efficient energy supply and helping utilities prepare for extreme weather events.

There is potential for AI to do much more, but there is significant risk too and a degree of regulation that reflects that risk. If energy supply or networks go down the consequences for society are calamitous.

Because of deep, wide digitisation everything, from defence to heating to information networks, security networks and banking, relies on reliable power. So regulators are rightly cautious about allowing insufficiently tested technologies into systems regulating the generation and distribution of power.

The EU’s AI Act which is considered the current gold standard of AI regulation, classifies AI systems used in critical infrastructure, such as energy, as high-risk. These systems are subject to stringent requirements to ensure safety, security, and reliability. While the Act does not outright prohibit the use of AI in these areas, it imposes significant regulatory obligations to mitigate risks. In time AI could be used with the very large systemic tasks such as balancing the grid to make sure supply and demand are evenly met and all generated energy is used efficiently. Or changing the way the network operates at different times to accommodate electric vehicle charging based on extra information harvested from the vehicles themselves.

Unlike the EU, not all national policy makers and regulators consider that there is a need to legislate specifically to regulate the use of AI in the energy sector. For example, in the UK the regulator, Ofgem, has prepared guidance for consultation which emphasises a pro-innovation approach, encouraging the safe, secure, fair, and sustainable use of AI while minimising regulatory burdens. Ofgem aims to support innovation by providing clear guidelines and good practices for stakeholders in the energy sector, focusing on risk management and ethical AI deployment, but it considers the current legislative and regulatory landscape to be sufficient to regulate the use of AI in the UK energy system. It will keep that position under review.

As the technology and regulation develops it may follow a similar pattern to the regulation of cybersecurity in the energy system, where regulators have increased their interest and scrutiny as technology has developed and the cyber-security threat to critical national infrastructure has increased. Regulators are now extremely vigilant about cybersecurity capabilities, safeguards and protections. As AI is deployed closer to the centre of energy systems the regulatory scrutiny of it will increase too because energy infrastructure is absolutely critical to the whole of society.

If EU AI regulation is regarded as most rigorous, it will be interesting to see whether, as in other areas such as data protection, the EU AI Act influences the approaches take in other jurisdictions to the regulation of AI Energy systems are heavily interconnected between the EU and neighbouring countries such as the UK  so alignment on standards and the regulation of AI within the energy system would be helpful to businesses who may own and operate energy assets in multiple jurisdictions

What will change the picture regarding the impact of AI is the amount and quality of data that’s available. Most homes and businesses now have smart meters which gather unprecedented intelligence about how energy is used, when and what for. And electric vehicles and the related vehicle charging infrastructure can supply similar levels of knowledge and insight about usage and charging patterns.

This could transform generation and supply by aligning them with usage more closely than ever before. Doing this would be enabled by the power, insight and control that AI can give. AI as a means to support better strategic and day to day decision making could have a major impact on energy markets, distribution and efficiency.

But to do this we need to find a way to share all this data effectively without breaching people’s privacy rights, compromising commercial confidentiality or breaking competition law governing co-operation between competitors.

This is a challenge, but it’s one that we face in other areas where data sharing is becoming possible through new data-sharing mechanisms, such as data trust frameworks. It’s a problem, but a solvable one.

So the future of AI and the energy industry is about closer co-operation and careful, well-regulated shifting of the technology from the periphery to the centre of energy system operations.