AI offers enormous potential for the energy industry and is a fundamental building block of the clean energy transition. It will lead to greater automation, saving money and driving operating efficiencies, the development of new products and services and - perhaps most importantly - enable a reduction in carbon emmissions. But the extent to which it will deliver the greatest possible returns for the energy industry will depend both on the acumen and involvement of senior executives to identify opportunities to create new value streams and ensure that projects progress from proof of concept stage, but also active and constructive engagement with regulators and policymakers.

 

Policymakers will have to walk a fine line between strong regulation to manage the known and as yet unknown risks of AI, while giving the young technology room to develop and providing the necessary incentives. And regulation on data access must protect consumer rights to privacy, which the Commission has recognised as a priority, but also respect companies’ legitimate interests and allow adequate access to data for solution providers. Somehow a compromise must be found that allows data to be anonymised, aggregated or otherwise sanitised so that it can be fed into digital interfaces without jeopardising consumer trust. Finding the ‘goldilocks’ level of regulation, ensuring the appropriate data is open and facilitating interoperability will be the key to success.

 

Narrow AI refers to the technology when it is able to handle just one particular task. General or strong AI is more complex. Applications of narrow AI will drive efficiencies in the short term, which will boost confidence in the technology. General AI projects will be more transformational but will take longer and will need to prove their maturity before scaling up. Complex decision-making required in the real world will need to be translated into AI language before we will be able to fully trust a machine to drive our cars, for example. But if trust can be established and ethical considerations are taken into account, the limits of what is possible with AI will likely be pushed to the boundaries of current human imagination.

Artificial Intelligence and other exponential technologies, such as digital communications, high speed internet (5G), blockchain, computing power and memory are becoming more and more interconnected. They will play a critical role in the clean energy transition and the unprecedented nexus of these technologies presents significant challenges and opportunities for the power sector.

 

Artificial intelligence is breaking through the limits of human capabilities at just the right time, as radical transformation of the electricity sector is essential for a cost-effective clean energy transition. The digitalisation and decentralisation of generation, transmission and distribution with real-time dispatching of flexible assets, interdependency and interoperability create an enormous data footprint that cannot be managed with traditional operating systems. But the climate emergency demands this crucial mission must be accomplished.

 

This paper which has been developed by the Eurelectric Beyond Digital Platform, will examine the opportunities for utilities to harness the benefits of AI and how this is paramount in order to deliver Europe’s Green Deal aspirations. It will also document some of the challenges and the latest regulatory steps seeking to address them.

Hover over the infographic in order to share it

 

While Europe has taken a leadership role in tackling the climate crisis, it has been slow off the starting block in the AI race, so far losing ground to the US and China. A new European Parliament special committee on Artificial Intelligence in a Digital Age (AIDA) wants to close the gap and see European companies developing digital talent and platforms, supported by greater investment in AI. Despite the COVID pandemic, Europe should still aim to attract more than €20bn/year of investment in AI over the next decade, EU Commissioner for the Digital Age Margrethe Vestager has said. The EU’s €750 billion green recovery package Next Generation EU should be a good source of funding for the digital transition as many AI projects will facilitate the adoption of distributed renewable energy.

 

Romanian MEP Dragoș Tudorache, chair of the AIDA committee, said: “It is commonplace to lament our place in the AI race. Clearly, we are behind in investment, by quite a sizeable margin too, but I do believe if we wake up soon enough, we can make up this gap.” However, the latest budget earmarked less than €500 million spread over seven years to strengthen the EU’s digital skills, which will not be adequate to deliver the boost it needs to catch up to the AI superpowers.

 

AI experts prefer the term ‘augmented intelligence’, highlighting the symbiotic future relationship between man and machine rather than the sci-fi image of a rogue android with a mind of its own. Human oversight will always be required at some level to ensure future ethical dilemmas are addressed appropriately and that the machine’s interpretation of data and algorithms makes sense in the real world. In fact, AI and other digital tools will enable the democratisation of energy away from centralised control desks, so that consumers will play a much more conscious role in how they use power.

 

Good data is like oxygen for this new man/machine hybrid – it cannot function without it. The success of AI applications in the energy sector will depend on access to widespread high-quality data and data infrastructure, which is now possible as data storage capacity has increased, while its cost has plummeted. But regulation and policies will have to balance the need for data sharing with the protection of consumer privacy and preserving Europe’s industrial confidentiality and competitiveness.

 

As Moore’s law that predicted the doubling of the number of transistors on a chip every two years is coming to an end, the next generation of computing efficiency will be found in software, algorithms and cloud applications. AI is already being used extensively in the energy industry for incremental efficiency improvements which deliver small cost savings on a wide scale. But the jury is still out on the extent to which it will create completely new business and revenue models. It will clearly allow the workforce to become more agile and dedicate a larger proportion of their time to customer-focused value-added activities rather than drudgework. Many repetitive jobs such as data entry, quality assurance or candidate screening will be automated.

Forecasts for the AI business opportunity are staggering. Gartner predicts that AI augmentation will create $2.9 trillion in business value globally next year, and 6.2 billion hours of worker productivity. The International Data Corporation predicts that worldwide revenues for the AI market including software, hardware and services will reach $156.5 billion in 2020, an increase of 12.3% year-on-year. AI revenue will top $300 billion by 2024 with a five-year compound annual growth rate (CAGR) of 17.1%, it says. If Europe continues at its current pace on AI it could add around €2.7 trillion, or 20% to its combined economic output by 2030, according to McKinsey. But if it accelerates to catch up with the US, a total of €3.6 trillion could be added to GDP in this period.

 

Nearly 25% more companies used AI last year compared to 2018, and most executives report a subsequent increase in revenue, with 44% of those who adopted it saying AI reduced costs, according to the latest McKinsey Global Survey. But early mover advantage will fade soon, warns the Deloitte AI Institute in its report “Thriving in the Pervasive Era of AI”. While using AI for efficiency improvements can boost productivity, companies can go beyond this objective by leveraging AI to create new products, the report says.

 

Time will tell whether AI lives up to its hype, but some predict it will lead to future innovations we cannot even imagine. Google’s CEO Sundar Pichai said that AI is one of the most important things humanity is working on, potentially more profound than electricity or fire. It is a good analogy, because while the benefits are huge, the implications of careless management or losing control are serious.

 

A careful balancing act will be required to manage the potential risks, avoid bias and safeguard consumer privacy while allowing sufficient access to data and designing regulations that are robust and effective without strangling innovation.

 

What is it?

 

AI is an umbrella term for a range of technologies which can simulate human reasoning powers by analysing data. At its core AI has the fundamental impact of reducing the cost of prediction and as that cost falls, then more and more activities will be shaped as prediction issues. In the same way photography moved from a chemistry challenge to a digital challenge, so many issues will move to being prediction issues. Whether self-learning models or algorithm-based management, machine learning, deep learning or reinforcement learning, utilities have a range of innovative business models and services to choose from.

 

Reinforcement learning can help optimise complex decisions, especially when good output examples are not available. In this way the model can learn sequentially which is the best action to take and this could often be applicable in the energy sector. It creates a decision strategy by testing multiple combinations and identifying the right one. Use cases include automated driving, financial trading, optimising engineering systems and energy management, and fine-tuning CAPEX decisions.

 

Machine learning and deep learning can be used to interpret patterns in real conditions and use them to make predictions, for example for power outage prediction, asset optimisation, fraud detection, predictive maintenance and load forecasting.

 

The digital twin is a cyber model of a physical asset which can be used for network flow management, equipment failure detection and asset management.

 

Blockchain, a distributed ledger technology associated with AI, can increase transparency and guarantee the origin of electricity.

 

 

AI can be divided into weak/narrow and strong/general categories. Narrow AI performs a single task or a set of closely related tasks such as weather apps or digital assistants. They tend to focus on finding efficiencies and can already be found in a surprising number of everyday applications. General or strong AI can handle a range of complex tasks such as data processing and decision-making.

 

Cloud platforms such as Microsoft Azure, Amazon AWS, Google Cloud and general-purpose AI frameworks such as IBM Watson provide AI services that are already available to many utilities. More bespoke services are offered by companies like C3.AI, depsys, OpenText or large utility-focused engineering firms such as ABB, GE or Siemens and systems integrators like Accenture, Cognizant, Infosys, Wipro and TCS. Many utilities are working with or have acquired AI start-ups, sometimes leading to a culture clash between the old and new style of working.

 

Working with start-ups presents significant opportunities but also some challenges, including technical, practical, legal, reputational and IP ownership risks, according to consulting firm Best Practice AI. Many of these risks stem from cultural gaps so it is imperative to understand them before utilities and start-ups work together. Utilities are starting to understand the need to ‘disrupt themselves’ with more agile, innovative thinkers, but they are conservative in nature for a good reason, as they cannot jeopardise security of supply if a project fails.

 

In the short term the immediate applications of AI are likely to be around optimising existing assets but the highest potential for AI application can often be found at the intersection of different sectors, for example the electrification of transport and heat, smart homes to maximise energy efficiency through integration and autonomous optimisation, flexibility services for grids and edge computing to enable the rapid response of decentralised energy assets with limited intervention. But while new AI projects may promise some ‘silver-bullet’ solutions, often it is at the point of integration with older legacy systems that obstacles are encountered.

 

A recent study by the French Association Think Smart Grids has shown that almost all European utility companies have tried Big Data projects through Proof of Concepts, often with start-ups. But Big Data has so far failed to live up to its expectations as the ‘next big thing’, according to AI specialist DC Brain in a recent paper.

 

Public funding for innovative projects is available, for example under the European Union’s Horizon Europe programme, which focuses on Europe-wide co-operation rather than deep tech development and niche products. The Commission has proposed that the Union allocates at least €1 billion per year in funding from the Horizon Europe and Digital Europe programmes to invest in AI. The Digital Europe programme will also help to make AI available to small and medium-size enterprises through digital innovation hubs, strengthened testing and experimentation facilities, data spaces and training programmes.

 

But there is a yawning chasm between pilot phase and commercialisation. Start-ups need to prove that they can actually see some adoption of the solutions they offer when the market is mature enough. Start-up incubators such as Free Electrons, Creative Destruction Lab Oxford AI programme, EIT InnoEnergy’s Business Booster and the Tech Boost for Smart Energy and Industry consortium led by French Alternative Energies and Atomic Energy Commission can play a useful role in matching up innovative projects with interested utilities which have the balance sheet to bridge the ‘valley of death’.

 

The issue of how AI is being applied to the world of energy has not gone unnoticed by the UK as it leaves the EU and seeks to find new relationships in the world based around research and innovation. On the one hand, it has a good track record of decarbonising and also has more AI start-ups than Paris and Berlin put together, however from this strong potential position, and despite the fact that it has recognised the 4th Industrial Revolution as important, it has failed so far to lay out a comprehensive AI strategy for energy. There is however work going on to create an International Centre for AI, Energy and Climate Change which is seeking to create an ecosystem that brings together policy and regulations, applied research, industry, an incubation space and funding, with the aim increasing the adoption of AI and creating new start-ups. “We believe that you need to create an ecosystem that reduces the barriers to entry for AI practitioners and allows hundreds of new start-ups to be created,” says Paul Massara, former CEO of Npower and co-founder of the Centre.