ESG Reporting with AI

The Green Key: AI and the future of sustainability reporting

In recent years, the accounting profession has come to be characterised by its conviction to swiftly adapt to the changing tides of technological development. With the dawn of the AI revolution, this principle remains unchanged, as the profession seeks to keep itself up to date with the latest trends, from super apps to generative AI, and from cloud to blockchain.

In order to take stock of these momentous changes, Santiago Bedoya-Pardo, Reporter at International Accounting Bulletin investigated the needs of the market regarding these emerging technologies, looking at reports issued by leading professional bodies and talking to co-founder and chief digital officer, Brian Sathianathan.  

The advent of these technologies, however, comes at a crucial point not only for the profession, but also for the world at large. As individuals, businesses, and governments across the globe face the looming threat poised by climate change, it is more relevant than ever for all parties to harness the strength offered by technology in the fight for a greener, more sustainable future.   

However, what role is the accountancy practice to play in this? What can accountants do to fight climate change? The answer is more obvious than most would come to think, a fact made explicit by a recent ACCA publication titled Sustainability reporting – the guide to preparation. The report, published prior to the momentous climate summit COP28, sets out key steps to help organisations start to prepare for sustainability reporting. It is when examining the question of sustainability reporting that the role to be played by accountants in the coming fight against climate change becomes explicitly clear.  

As stated by PwC, the ESG reporting landscape is both complex and changing at pace. With the coming of new regulatory frameworks, together with rising levels of stakeholder scrutiny, a magnified demand for transparency has been created. Those accessing the services of accounting firms across the globe are no longer concerned exclusively by their own profit and success, but also by the commitment to safeguard our planet. However, how can organisations rise to the challenge? The answer is in the realm of data and technology.  

As organisations and firms are expected to adapt to this new coming of age of technology and climate action, so are those in decision-making roles across the world. CFOs, once responsible for cashflow oversight, financial strategising, and internal reporting, are now in for a change. As organisations seek to demonstrate their willingness to adapt and take the necessary paths to tackle the climate challenge, CFOs are becoming increasingly expected to lead the way in relation to sustainability and ESG, especially by means of reporting.

Read the report

Brian Sathianathan co-founder and chief digital officer

The challenges

The task at hand, however, faces a myriad of challenges given the constantly evolving global landscape. Amongst these challenges, however, that of data collection stands out in the quest to meet ESG compliance requirements. The data needed for ESG purposes is, as stated by Sathianathan, “found in many different sources, different document types, and in various formats all following different methodologies and at different verification levels. All the data pieces can be a bit rough to define at times.” 

It is in light of this that AI now emerges as an extremely powerful solution for overcoming these challenges, allowing for organisations to have access to a smooth and seamless ESG data collection process. Common forms of AI accessible to the general public, such as generative AI, are starting to become increasingly good at carrying out such tasks, which have often been dealt with by means of ‘traditional’ AI applications.  

Commenting on this, Sathianathan said: “In the past couple of years we have seen AI develop and advance at a very rapid rate. We’re going at a speed of 3000 times what we expected, that’s what we have seen over the past two years. Even over the past six to eight weeks we have seen massive leaps in terms of what these technologies can do. Almost every morning when we walk into our lab we come across a new discovery. 

“It is thanks to these developments that access to such tools is playing an active role in changing the way companies keep track of, measure, and assess their performance against ESG” added Sathianathan.

However, the issue of data gathering is not the only factor leading organisations to adopt new AI-driven strategies to face the climate challenge ahead. As noted by several international organisations, including Baker Tilly International, the complexity and scale of the ESG regulatory framework, together with a myriad of new sustainability ratings continues to evolve and develop in real time.  

In recent times, this has been particularly true when assessing the European Union’s introduction of the Corporate Sustainability Reporting Directive (CSRD). Published in 2022, the aim of the CSRD is to provide a unified framework for corporate ESG reporting. It aims to increase transparency, comparability between companies, and the reliability of published data, gradually bringing sustainability reporting up to the standards currently required of financial reporting. It will require companies to publish a sustainability report with information on the governance, strategy, impact management, objectives and metrics of ESG information, which means companies will need to look at how they structure their ESG governance, strategy and data collection systems. 

According to Baker Tilly International’s ESG lead, Arnaud Bergero, “The CSRD will completely redefine the corporate ESG landscape and there are still too many companies that are unprepared for its implications despite the implementation of the directive being only a month away. However, companies are not starting from scratch, and can leverage existing ESG processes and politics to address the strategic challenges arising from this new obligation.” 

These changes, however, are not exclusive to the European front. When looking at south Asia, India requires its top 1000 listed companies to disclose ESG data according to what is required by the country’s Business Responsibility and Sustainability Reporting frameworks. In the United States, the newly proposed greenhouse-gas emissions reporting requirements, proposed by the Securities and Exchange commission, will create a requirement to report on Scope 3 emissions (emanating from third-party suppliers and customers) across the board.  

The evolution of these regulatory frameworks will further drive the need to obtain and analyse increasingly complex data sets across the world, further creating incentives for global organisations to adopt the cost-effective alternatives proposed by AI technologies in order to meet their ESG reporting requirements. 

Source: GlobalData

Unsuprisingly, several accounting companies are jumping on the Gen AI bandwagon. KPMG is adopting Gen AI as part of an in-house system that aims to improve efficiency and insights. PwC is also on a Gen AI mission and will invest $1 billion to improve its Gen AI presence in the US within the next three years. The company has also partnered with both Microsoft and OpenAI to automate tasks and increase overall efficiency.

How can AI tackle these challenges?

IAB’s conversation with Sathianathan soon came to focus on the different ways in which AI developers across the world, including, are actively seeking to develop the right tools for clients around the globe, clients who are not exclusively seeking to maximise their utilities, but furthermore take the right actions against climate change.  

According to Sathianathan, in order to fully grasp the ways in which these tools can be of use, it is first key to take into account the history of AI development. He said: “Back in the 90’s, we had AI setups that were very much statistical. They were good at statistics-oriented decision making and problem solving. Many different organisations made use of these models, especially banks at the time.  

“However, between 2007 and 2009, you had new additions come up in the field, such as deep learning, the use of neural networks. We used these increasingly efficient tools to enhance the ways in which we were teaching these systems. However, by 2018, we saw the rise of the ‘transformer’ models.  

“These new models were not only capable of learning and using problem-solving skills, but also of retaining knowledge. Back in the day, early large language models (LLMs) would be able to consume information, such as books, think of the Harry Potter series, but by the time they were done with the first chapter and were ready to move on with the second one, they would’ve already started forgetting the opening lines. They did not retain information. 

“However, with this new ‘transformer’ model, these LLMs are now capable of defining which bits of information are critical – they know what to pay attention to, and what to remember.” 

According to Sathianathan, the mathematical skills which characterised the AI models of the 90’s have remained in place, while the language and critical information handling skills developed in the late 2010’s have now joined forces to deliver far more effective AI tools for users globally. And their uses are not limited to day-to-day tasks. Based on the findings made by his team at the lab, these new AI models are due to play a crucial role in the fight against climate change in the future.  

According to the findings made by Sathianathan and his team, not only will these new tools facilitate CFOs in the development and execution of sustainability reports for internal and external use, but they will also come to play a crucial, practical role in terms of resource optimisation globally. On this, Sathianathan said: “These models are not only becoming more and more effective at sorting out large data sets, the types of data sets accountants and CFOs have to deal with every day, but they will also allow companies to best distribute and allocate resources. AI models are now able to keep track of temperature, food consumption, and many other key elements that make up the core of how organisations should be practically structuring their sustainability strategies.” 

What the future holds

For Sathianathan, the era of AI technology supporting the battle against climate change is only getting started. The rapid and constant development of LLMs that can be actively fed information and given precise instructions will not only facilitate the gathering of data for ESG reporting, but it will further allow for the different data points employed to be crosschecked and verified, actively contributing to the quality of these reports.  

However, given the current state of the world in relation to climate change, many would wonder – isn’t it all too little to late? For Sathianathan, it’s “safe to assume that pessimists will be right today, but that we can prove them wrong tomorrow.” 

He further said: “Even now we can ensure that these AI models will not only gather this data, but further create dashboards allowing for CFOs, accountants, and other financial services professionals to actively monitor the data being gathered in real time. Tasks that once would’ve taken a week are now being dealt with in an instant.” 

To Sathianathan, the advances made by these technologies are a reason for hope in the months, years, and decades to come. When once they would only have been able to process up to 2000 words at a time, they can now be fed pieces of information reaching the hundreds of thousands. Thanks to organised, global efforts, such as those being enacted by the UN Policy Forum, it is only a matter of time before the very technology now used in the accounting profession for reporting ends will be widely used across fields such as agriculture, energy and food security, amongst others.  

“You want those who believe that it is all over to be wrong”, Sathianathan concluded, “We, however, have the tools at hand to prove them wrong.” 

Main image: Shabnam Pervez (MBPsS) Thematic analyst in GlobalData's London office