The energy transition can benefit from the introduction of Artificial Intelligence but it has work to do to gain the sector’s trust and ensure safe operations.
Artificial intelligence (AI) can already stake a good claim to be the phrase of 2023. It is now a well-established household name, in no small part due to concerns about the potential for the technology to generate thoughts of its own, replace jobs and spread disinformation. Such worries are certainly not without merit, and they broadly represent the hopes and fears of many.
Nevertheless, the emergence of AI comes at something of an opportune moment as humanity grapples with one of its most pressing challenges – climate change. The end goal is clear – to slash carbon emissions and reach net zero. But there needs to be a constant within that: an affordable and secure energy system.
Our Energy Transition Outlook UK 2024 highlights that electricity demand in the UK will increase by a factor of 2.3 by 2050 compared to today. This demonstrates the significant scaling of the nation’s energy system by 2050. The pressure to digitally overhaul the sector to guard against recurring challenges is mounting, and that depends on innovative, disruptive technology.
As such we find ourselves at a critical juncture, where we must look at the remarkable potential of AI and consider its integration into our energy systems so it may be a boon, rather than a peril. The energy transition demands it.
A whitepaper to light the way
In becoming the highest trending Google search globally over the past year, responsive chatbot system, ChatGPT, has opened our eyes to the wider possibilities of the technology. Within the energy space, AI has the potential to unpick many of the complexities arising from the transition to a low carbon energy system.
To better understand how AI can be harnessed as an asset, DNV, a global risk management and assurance provider recently published its ‘AI Insights: Rising to the Challenge Across the UK Energy System’ whitepaper. Launched to coincide with the SPE Offshore Europe 2023 conference in Aberdeen, it features contributions from leading organizations and individuals at the forefront of understanding and implementing AI-technologies for the sector.
Barriers to the use of AI
Understandably there is still a degree of trepidation in the energy sector about the adoption of AI, and a degree of healthy scepticism can act as a good safeguard. Having learned safety lessons the hard way, the industry in general has a natural aversion to risk, and while this is generally the right approach, it can act as barrier to new ways of working.
In our conversations, we found that there is a culture element around the discipline of engineering, one that has an aversion to risk and a low tolerance to error. Key to AI machine learning models is that data science involves measuring test error and being explicit about it.
Subsequently, when this then comes to the adoption of new technologies, the flagging of an error margin can have a negative impact on the receptiveness of companies. When compared to current ways of working and manual work by engineers, where there is perhaps no way of measuring performance, new algorithms can be held to a higher standard than the existing process.
There is a consensus that data and cultural barriers are also slowing project owners and developers’ adoption of innovative technology, to the detriment of the energy transition. There are economic factors to consider too. In our research, we found that companies have struggled to establish full cost benefits or understand what the savings could be achieved by embracing digital technologies.
Then there is the skills gap that is hampering the growth and development of much of the energy system now. There are a limited number of people who understand both AI and the complexities of the energy sector, let alone how the two fit together. Bridging that gap need not be difficult, but it will require communication and, most importantly, time, the latter of which is often at a premium.
It may sound like a simple solution to a complex problem, but collaboration is key in developing a more rounded knowledge of energy systems and AI’s place within them. Sharing knowledge and data will only aid our understanding of dense challenges, and while there is a new willingness within the sector to work together, there is certainly room for improvement.
Overcoming unease and building trust
Effectively harnessing AI's potential to support the UK's transition hinges on a concerted investment of both thought and time. There is a myriad of obstacles - whether technical, cultural, or somewhere in between – to overcome, but they all have one common root: a lack of trust.
DNV’s ‘Assurance in the Digital Age’ report found that trust gaps emerge and widen as the development of technology outstrips regulation and standardization. Building faith is a prerequisite for unlocking the advantages of AI, and that in turn requires responsible use of this great tool. To access the potential value there is a need for a comprehensive trust framework that encompasses technology, users, and the environment.
Providing assurance has been DNV’s core business since 1864, so we have a big part to play here in building trust in AI systems by evidencing, assessing, and evaluating the technology and its uses. The philosophies and processes to assure technologies are well established, and good practice in software development is also mature. However, AI systems present unique challenges and are inherently opaque in operation. An independent voice is essential to build a strong bed of trust.
Regulation and independent assurance are the primary instruments by which we establish trust in the energy system and the technologies that underpin it. For example, health and safety in the oil and gas industry is a clear-cut example of this. In the wake of the 1988 Piper Alpha oil platform explosion in the North Sea, which claimed the lives of 167 men, the UK Offshore Safety Case Regulations were implemented, prompting a step-change in offshore safety. A critical aspect of these regulations is the requirement for independent verification: establishing trust through oversight.
Typically, regulation has been reactive rather than proactive, and governments across the world are already playing catch-up with AI technology. As digital technologies spreads across borders, it will become increasingly hard to regulate, meaning efforts to build trust becomes fragmented. This is changing though and guidance designed to shape the use of the burgeoning technology in critical industries is taking shape. The AI Safety Summit – held in the UK in early November 2023 – resulted in the "Bletchley Declaration" aimed at boosting global efforts to cooperate on artificial intelligence (AI) safety. The declaration, by 28 countries, including the United States, China, and members of the European Union, was published on the opening day of the Summit.
The UK Government – through the Department of Science, Innovation and Technology (DSIT) – openly encourages the development of a flourishing AI assurance ecosystem. It’s acknowledged as an important aspect of broader AI governance, providing the basis for industry to confidently invest in new products and services via a framework for effective monitoring and compliance through regulators. DNV’s own framework – a recommended practice for the assurance of industry AI systems – is listed as a credible approach via a case study on the UK Government website.
The end goal is clear, the path less so
Industry stakeholders now believe that AI has the potential to make a significant contribution to our future energy system, highlighting possible solutions by painting a fuller picture. Identifying an ideal path to said solutions is still up for debate though, and differing schools of thought are emerging. While the best route remains unclear, inspiration can provide the required first steps, and there are several examples of AI applications in the energy system that should give us heart. One key area is around the electricity grid, where data driven decision making can help to balance the intermittency of renewable sources like wind and solar by mapping supply and demand.
Within the underlying technology stacks, the velocity of advanced analytics solutions and AI algorithms used will need to increase to analyse the vast amounts of data generated by the grid. For example, predictive analytics can help anticipate equipment failures, optimize grid operations, and improve energy forecasting. The quantity of data held in the energy sector is already vast and the technologies previously outlined are increasing that volume exponentially. Being able to process those large datasets and derive insights from the industry as a collective will bring significant benefits. A big picture, whole systems perspective is essential and technological innovation will characterize the energy transition.
Whilst the power grid provides candidate use cases for the AI, the end goal is clear – to slash carbon emissions and reach net zero. But there needs to be a constant within that: an affordable and secure energy system powered by digital technologies.
As energy value chains become increasingly connected, secure and trusted connectivity is key and one that will grow over the coming year as investment in digital technologies increase. Standardization across the industry comes from increased interaction and cooperation. This implies open data and data sharing, and the need for information to be in an agreed common format to facilitate data exchange across the supply chain. Building AI models on standardized libraries is the required foundation of having safe, effective, and efficient solutions. Anchoring those principles to highly performant technologies will in turn enable a safe, secure, and sustainable energy system.
Written by Graham Faiz CEng CITP FBCS, Head of Digital Energy, DNV