Use of AI in sustainability reporting - Part 2: Application tools
Sustainability reports are still expected from many companies even though the requirements for these are being significantly reduced as part of the respective efforts by the EU (the keyword here being the Omnibus Regulation). In the first part of our series, last month, we guided you through the sequence of steps in the report preparation process. In this issue, we describe how AI applications can be used as aids in the individual process steps. In doing so, we present specific AI tools by way of example.
1. AI tools for the materiality assessment
Drawing up an IRO long list
Here, ChatGPT can serve as the first step in the materiality assessment by preparing a long list of relevant sustainability aspects (IROs). By writing a precisely worded AI prompt it is possible to identify the industry-specific sustainability issues. On the basis of the information made available by the user on the industry, the stakeholders and the business model, it is possible to assess the sustainability issues in respect of financial risks and opportunities as well as the impacts on people and the environment. This method offers an introduction to the materiality assessment and provides an extensive long list that can serve as a foundation for the further analysis.
Climate scenario analysis
The aim here is to make transparent the potential impacts of climate change on the business in order to be able to assess these subsequently. A tool, such as, for example EarthScan, would be able to provide an analysis of the climate risks threatening your business assets by using data modelling, machine learning and climate science. The tool takes into account various time horizons and scenarios and generates reports in accordance with the disclosure requirements of the Corporate Sustainability Reporting Directive (CSRD) and the Task Force on Climate-related Financial Disclosures (TCFD).
Comprehensive solutions for performing a double materiality assessment
Besides specialised tools for specific aspects of the materiality assessment, there are also comprehensive solutions that cover the entire double materiality assessment process. Tools such as Double Materiality Assessment (DMA) Generator from denxpert, or Materiality Master from the eponymous provider, support companies in performing a double materiality assessment. With the DMA generator, users first enter the name and primary location of the company. The AI then assesses the company’s sustainability-related impacts, risks and opportunities (IROs) on the basis of publicly available information and open source research databases. According to denxpert, the requirements and recommendations of the ESRS and EFRAG’s Materiality Assessment Implementation Guidance are taken into account here. Finally, a comprehensive DMA report is generated, with suggestions for IROs and sent via e-mail.
2. AI tools for the collection and evaluation of data
Functions
Most AI tools for the collection and evaluation of data in the ESG sphere provide a range of basic functions that support companies in the analysis and reporting of ESG information. These tools automatically collect data that is relevant for ESG from different sources, then analyse and categorise them. In doing so, patterns, anomalies and errors are identified. Moreover, the systems automatically assign the correct emissions factors in order to calculate the CO2 emissions. The results are presented in clearly laid out dashboards, graphics and reports that help companies to identify trends and risks and to prepare standards-compliant ESG reports.
Examples
KEY ESG connects with a company's systems in order to collect and check data that is relevant for ESG from different areas, such as, finance, supply chains and environmental. The tool uses AI to verify data and to ensure that the calculations are accurate. The carbon accounting is in line with global standards, such as, the GHG Protocol and the IEA whereby automated emissions calculations for Scopes 1, 2, and 3 are made available. Furthermore, KEY ESG provides support for the preparation of reports that comply with the current regulatory requirements and continuously adapt to new standards, such as the CSRD.
Briink can support small and medium-sized enterprises that have to fill out extensive ESG questionnaires for customers or suppliers. Briink uses AI to automatically process ESG questionnaires by extracting the relevant information and metrics from existing company documents, such as, guidelines, annual financial statements or other internal documents. The AI is specifically trained on ESG issues.
3. AI tools for the preparation of the report
One example of this is the software company Code Gaia whose software uses AI to support the preparation of sustainability reports. The user has the option of uploading a document, such as, a corporate guideline for example, or entering context information in the form of bullet points or notes that are then analysed. On the basis of the analysed information and the requirements of the ESRS (European Sustainability Reporting Standards) the AI prepares a first draft text for the required publication. In doing so, the AI helps to formulate clear and well-structured sentences and provides support to users with the adapting and reviewing of the generated text in order to ensure that it is correct, company-specific and conforms to the ESRS.
ChatGPT can likewise be used here to translate complex data into understandable and target group-oriented texts. It generates draft texts, optimises the wording and ensures that the relevant metrics are presented clearly and concisely. Moreover, ChatGPT helps with the structuring of the report and provides, for example, tables of contents or suggestions for how to organise the individual chapters.
4. The limits of AI in sustainability reporting
Checking and reviewing
While AI models are able to identify patterns and formulate contents they lack a deep understanding of the context. Moreover, without relevant input data it is possible that the will AI generate plausible but nevertheless false information.
Regulatory challenges and legal risks
The sustainability reporting guidelines are subject to constant change. Obsolete or inaccurate data could lead not just to incorrect analyses, but also to regulatory infringements. Companies must therefore ensure that the AI systems deployed are regularly updated and conform to the current standards.
The limits of qualitative analysis
AI models are particularly efficient in the processing of structured quantitative data. However, they hit their limits when it comes to the interpretation of qualitative or context-related information. Issues that require deeper engagement with the content or human judgement thus cannot be fully automatically analysed.
Quality of the underlying data
The data and their quality are the key factor for accurate and reliable analyses. Complex reporting requirements, such as Scope 3 emissions that rely on external data sources, are especially challenging. It is therefore essential to establish robust data management processes and to ensure that all the relevant stakeholders - both internal as well as external - provide reliable and up-to-date information.
Summary and outlook
Integrating AI into ESG reporting enables companies to manage their sustainability reporting more efficiently. AI tools are however only able to provide support in this respect. At present, there are not yet any universal tools applicable as a whole that, almost at the touch of a button, are able to access externally available data and internal information in order to generate sustainability reports from them, especially since changes in the legal requirements and in the actual facts are still even faster than AI.
