AI and energy

Answer

In researching this topic, I came across a study that found that 93% of the environmental UN Sustainable Development Goals (SDGs) can be achieved with the aid of next generation digital technologies, specifically, artificial intelligence (AI). With further research, it’s become clear to me that not only can AI overwhelmingly power (excuse the pun) the SDGs, but if AI and AI-powered solutions are adopted quickly and on large scales within the future energy space, they could also be instrumental in our transition to a low carbon economy.

Smart grids

One way in which AI is likely to continue to affect the energy transition is by improving the efficiency and effectiveness of renewable energy sources by optimising their use. Smart grid management has the potential to improve grid agility and resilience as the AI algorithms can balance energy supply and demand and optimise transmission and distribution of power from sources such as solar or wind. Internet of Things (IoT) sensors across a smart grid can also help to detect risk at a much earlier stage and redistribute energy to balance loads and reduce outages more efficiently than the traditional grid infrastructure. As extreme climate events become more common, traditional grid operators are being forced to review current practices and use innovation to fulfil the demand for renewable energy while retaining operational efficiency.

Conveniently, smart grids also offer consumers an insight into the energy they use and, any energy they produce and store themselves. Representing something of a paradigm shift from traditional energy generation, the rise of the ‘prosumer’, that’s, a consumer that produces energy themselves either to offset their own energy demand (and costs) or to sell back to the grid as surplus, has the potential to accelerate the success of smart grids. Data arising from the smart grid can also increase sustainability and transparency by engaging consumers and helping to democratise the grid. 

However, smart grids can come with significant initial overheads and time demands. Equally, smart grids can be vulnerable to cyberattacks given the intricacy of AI technologies, sensors and software involved in transmitting and distributing renewable power. In discussing how developers could overcome these challenges with an associate in the corporate future energy team, I found out about a successful grassroots initiative, Brooklyn Microgrid, which started with peer-to-peer energy transactions and has now become a marketplace for residential and commercial prosumers to buy and sell locally generated renewable energy. On further research, I discovered that ‘Exergy’ is a permissioned data platform used across the Microgrid, which is powered by blockchain technology to create localised energy marketplaces for buying and selling energy across existing grid infrastructure. As blockchain technology is powered by a decentralised distributed ledger system whereby transactions are recorded and verified by the prosumers themselves, it’s less vulnerable to cyberattacks (albeit no technology is ever 100% secure). Perhaps a model interlinking AI and blockchain technologies to handle data to produce and transmit renewable energy would offer a more sustainable network.

Natural language processing (NLP)

ChatGPT hit 100 million users in February this year, three months after launching and surpassed one million users in just five days. By comparison, TikTok took nine months to reach 100 million users, Instagram took over two years and Facebook took four years. The technology is clearly powerful and its uptake has transformational abilities. In its current form, ChatGPT uses NLP and machine learning (ML) algorithms to offer a conversational tool that can provide bespoke information to users. According to the Harvard Business Review, AI is even thought to be better than humans at data-driven decision making. Against the backdrop of the energy transition, ChatGPT can be used to analyse large datasets, optimise energy demand and help to support forecasting. NLP technology may then provide a framework for decision makers to identify solutions to overcome certain challenges. In organisations or public sectors that struggle to make relevant investment decisions relating to energy, a technology that can assist in developing a clear roadmap with solutions may prove invaluable to ensure greater clean energy security and equity. 

However, given the energy intensity of the application itself, organisations using ChatGPT must be alert to the amount of energy used by software algorithms and hardware to ensure that it doesn’t add to the problem that it’s trying to resolve. In any case, the technology is still in its infancy and it only repeats information that it’s already seen before. The lack of availability of data is a challenge that’s mirrored across all AI technologies; however, with ChatGPT, the lack of availability of data presents the unique challenge of inaccuracies without knowing where the inaccuracies lie. ChatGPT may not be able to lead us to net zero alone – we’ll also need clear regulations and structures if secure, accessible and standardised data sharing is going to provide a collaborative clean energy future.

Google as a case study

I recently had the opportunity in my seat to attend the Solar and Storage Live UK 2023 conference in Birmingham. During one of the sessions that I attended, Google spoke out about its goal to help to accelerate the transition to clean solar energy. In developing its Solar API tool, Google is providing residential level data across multiple data sets, covering more than 320 million properties in 40 countries. This piece of AI-powered technology blends ML with satellite imaging to deliver a comprehensive solar data solution that can be used by:

• individuals or companies to assess the benefits of installing solar across different configurations; and
• developers in providing the detail needed to create custom solar proposals.

By offering such clean technology across various levels, Google was one of more than 250 exhibitors from across the globe at the conference leading the way to reduce costs, increase efficiencies and inspire others to transition to solar power.

Google also enlightened us with ‘Project Green Light,’ another AI tool that it’s developed, which is relatively low cost and utilises existing infrastructure. Here, it’s using AI technology and Google Maps to model traffic patterns and make recommendations for optimising the existing traffic light plans. The model uses various data sets to assess what the most efficient way of timing traffic is for any given intersection. What I found particularly impressive was that it does not only optimise one intersection, but it can also be coordinated across a city to create waves of green lights to increase free-flowing traffic and reduce stop-and-go emissions by 10%. The project is now operational across 12 cities from Manchester to Jakarta and is thought to lower emissions for up to 30 million cars per month. Perhaps a more functional use of clean tech for cities with less access to emerging low-carbon technologies, the continued uptake of the project may significantly assist in meeting emissions reductions targets globally. 

ML advancements are offering powerful algorithms that allow data to be processed in new ways to identify efficiencies. At the conference, I was interested to learn that Google has been using technology initially developed by its subsidiary, DeepMind, to predict wind power output 36 hours in advance since 2019. Wind power is notoriously unreliable due to its variability and unpredictability but this technology allows energy providers to schedule grid inputs more accurately and better dispatch energy onto the grid. I learnt that Google Cloud is now developing this technology further to produce more accurate data and predictions of wind power production to assist those with wind farms to make smarter, more efficient and sustainable decisions to meet consumer demand.

Conclusion

Ahead of the UN’s COP 28, which is due to kick off in Dubai at the end of the month, the examples above show that there’s certainly a role for AI, as well as other technologies, as an enabler in bringing about greater optimisation and efficiencies to meet net zero. However, AI and low-carbon technologies are far from being the solution to the energy transition. Rather, I think that we’ll need to combine these with investments in financial, political and social movements towards a more sustainable, low-carbon future that benefits both businesses and consumers. With the International Energy Agency currently estimating that 50% of the technologies we need to achieve net zero by 2050 still in prototype phase, innovation is key to unlocking potential. The challenge at the intersection of future energy and AI lies in developing these technologies securely to maximise their effectiveness long term.

Emily Andrews is a trainee in the corporate future energy team at TLT LLP.