Artificial Intelligence and the role for ESG in real estate
Artificial Intelligence (AI) has been thrust into the spotlight, but what role could AI play in real estate when looking through an ESG lens?
9 minutes to read
The possibilities of AI, and more importantly, Generative AI, have spurred conversations across industries, and real estate is no exception. From better reporting to more efficient buildings, there are some critical considerations for the ESG space that will be increasingly pertinent.
"Poised to unleash the next wave of productivity"', Generative AI could add between $2.6 trillion and $4.4 trillion of economic value annually, according to research by McKinsey. Generative AI creates new data based on recognising patterns and strictures within existing data.
In contrast, traditional AI is used to analyse data and make predictions – Forbes describes the difference as "traditional AI excels at pattern recognition, while generative AI excels at pattern creation."
How can ESG in real estate benefit from Artificial Intelligence?
Whilst AI is utilised across industries, including real estate, and has been for a number of years, the democratisation of AI through the advent of ChatGPT and Bard, as examples, is opening up new possibilities.
There are significant opportunities for real estate and ESG to benefit from these advancements in traditional and Generative AI.
Below we explore some of its potential uses and implementations.
Optimising building use, assisting transactions and certification
Optimising building use will be critical for owners and occupiers to reach net zero goals and reduce costs.
Evaluation of building utilisation and sustainability has been implementing AI techniques for a few years, but the adoption and application is growing.
By attaching a device to the building management system (BMS), AI can analyse occupancy, energy consumption, waste and water usage data to provide insights for ESG management. This implementation can optimise energy consumption by analysing patterns, identifying inefficiencies and suggesting energy-saving measures which will reduce costs, waste and emissions.
For example, the UK's second-largest REIT, LandSec, began trialling AI in their headquarters in London in the first half of 2023. The trial will allow for two to three months of the AI' learning' to ascertain when people use the building, what time they enter/leave, which areas are occupied at what times of day, and how all this affects metrics such as temperature. After this period, the technology will work out the optimal time for heating and cooling and airflow systems.
The possibilities with optimisation enhanced, monitored and managed by AI, allows for more rapid changes if the building use changes and could be crucial in ensuring that net zero-enabled buildings achieve their designed efficiency levels.
Furthermore, the requirement to collate and present actual energy data is growing more prominent for property owners looking to sell assets. We surveyed property market investors about how ESG is impacting their investment decisions.
We found that ESG Reports, such as EPC Plus/Pathway reports and EU Taxonomy (which defines if an activity can be classed as sustainable) compliance are increasingly being factored into the due diligence process for acquisitions. AI could help gather the data and increase transparency in transactions. My ESG newsletter will dig deeper when the report is released.
Another aspect is that building certifications, such as BREEAM or NABERS, could see increased adoption if AI is used to assist in data gathering and validation.
NABERS UK is relatively new, many already own or are developing NABERS-rated buildings, and therefore we expect its uptake to grow, especially given the focus on building performance.
NABERS was originally developed in Australia and measures the energy efficiency of a building over 12 months, awarding a rating from one to six stars (six being the best) by comparing energy consumption to averages across the sector.
Data collection for reporting potential
Companies and entities are subject to more ESG reporting requirements, both from a regulatory standpoint and from stakeholders.
Regardless of which standard of reporting framework that is required, AI could streamline the collection of metrics, reduce the cost of doing so and help to standardise reporting
According to estimates, up to 90% of the data generated globally every day is unstructured – documents, images, videos, audio and social media for example – and non-uniform. In the differing ESG disclosures, several consistent metrics are to be gathered where the benefits of using AI to comb through unstructured data to identify and gather relevant data is clear.
Natural Language Processing (NLP)
One way this is possible is through using NLP. It is defined as applying computational techniques to analysing and synthesising natural language and speech – simply being able to go through various mediums and draw conclusions based on content.
In the real estate reporting context, NLP could help identify sources of emissions that fall within various scopes (1, 2 or 3) or other sustainability data that may be spread across unstandardised reports and documents, as well as from a broad range of stakeholders.
A different use would be to assess real estate's physical risk, i.e., climate change's potential impact on assets, which is not only becoming an increasing aspect of finance but is a requirement under the EU's Corporate Sustainability Reporting Directive (CSDR). This is useful, and in some respects mandatory, for investors and lenders to assess overall exposure. AI can gather relevant information related to various climate scenarios and the established impacts of these with applicable research.
In addition, there is the potential to enhance the accuracy and reliability of ESG data. With growing concern over fears of 'greenwashing', the enhanced data collection methods and standardised metrics presentation could help alleviate some of this hesitancy and allow for more comparability.
In addition, NLP can process millions of unstructured document sources, such as ESG reports and assessments, picking out evidence to either support/rebuke sustainability claims.
These examples and other uses of AI could reduce the current sizeable cost of complying with regulatory requirements.
Task Force for Climate-related Financial Disclosures (TCFD)
In the UK, the Task Force for Climate-related Financial Disclosures (TCFD) became mandatory in April 2022 for an estimated 1,300 firms.
Before implementation, the FCA conducted a cost-benefit analysis which estimated that for asset managers, the average one-off and ongoing costs for an individual large firm were £1.7 million and £1.2 million, respectively; and for an individual medium-sized firm, these costs were £1.4 million and £0.7 million respectively.
For asset owners, the average one-off cost for an individual large firm was estimated to be £1 million, with ongoing annual costs of £400,000; for a medium-sized firm, these costs were estimated to be £800,000 and £200,000, respectively. Therefore, the possibility of cost reductions could be significant.
Measuring the Social side of real estate
Seismic steps have been taken to quantify the 'E' or Environment aspects more easily, yet the 'S' or Social side can often be overlooked because of the inability to agree on what to measure and how to measure.
Firstly, relating to the governance element mentioned below, AI could identify best practices in terms of what to measure through analysis of questions, requirements, company reports and many other channels.
On how to measure, through the use of AI, social media sentiment and NLP, data can be amassed to measure social value and impact for the benefit of ESG investors.
AI algorithms can evaluate real estate assets and development projects' social and community impact by analysing data sources and providing insights into community needs, expectations, and concerns. This could be used to mitigate potential adverse effects and develop and enhance positive aspects.
In addition, effective communication strategies facilitated by AI algorithms build trust with community members and ensure their involvement in decision-making.
Artificial intelligence in Governance
The final letter of ESG that often gets less airtime is the 'G', or Governance. What does the 'G' signify and how will AI be deployed in this space?
The term governance often gets confused or oversimplified as running a company well or simply reporting structures, it is more than that.
The 'G' in ESG stands for decision-making governance considerations – S&P highlight some of the core elements as; the corporation's purpose, the role and makeup of boards of directors, compensation structures, shareholder rights and how corporate performance is measured.
One potential role for AI in the G is assessing what to report on and how. Property investors are increasingly needing to assess an asset on alignment with SFDR regulations and for EU Taxonomy compliance. Yet this can involve substantial data collection.
Regulatory and/or disclosure requirements often include many metrics and criteria for companies to consider.
For example, in the EU Taxonomy the technical screen criteria for construction and real estate has a significant number of elements to assess depending on the activity (i.e. acquisition of a building, renovation, construction etc.) and in terms of what constitutes as 'substantial contribution' and 'do no significant harm' for climate change mitigation and adaption.
These are the first two climate and environmental objectives; once the four other objectives have set criteria there will be an increased number of elements to assess. AI could help sift through and surface the relevant aspects and criteria.
Further, by leveraging AI technology, real estate investors and developers can ensure they align with industry standards and peers.
For example, with voluntary disclosures or ESG commitments, there can be an assessment of other entities and stakeholder questions/requirements to help alignment across the industry on metrics to use, how ambitious to be and how to measure.
We cannot ignore the risks
Good Governance is vital with a real need to have structures and processes in place to ensure that any data and uses for AI is robust and verifiable.
There are concerns that AI is unable to distinguish between real and falsified information, which could present risks on reliability and accuracy. Generative AI is trained on data sets up to a point in time in the past. They are unable to provide information in the context of recent events.
There are risks of falsified information. Much of the training data will be from a wide range of internet sources, including social media, which are known to be used to publish and promote false information.
There is also scope to manipulate or create AI visualisations and documents, which present a significant risk.
Similarly, Generative AI using large language models (LLMs) have been known to ‘hallucinate’ even when fed all accurate information. This is because they generate context based on the most likely pattern which may not always be right in the context.
Although through the different iterations of even ChatGPT this ‘hallucination’ has become less frequent.
Artificial intelligence bias
Training data will contain opinion which is likely to be biased. Generative AI solutions are not yet able to filter out bias and misinformation, meaning that the output may reflect, at times subtly, views that are not accurate or appropriate.
In addition, the ownership of data made available to AI applications and data ownership rights must be considered and will be of great debate and scrutiny.
We know that there is a lot more to come. Further advancements in the technological space and regulations will ensure that it is not manipulated and exploited. Yet there are clear and positive uses for real estate actors, which could ultimately lead to lower costs and emissions.
Subscribe for more
Subscribe below for a monthly outlook on all things relating to ESG investment criteria and the property sector.
Subscribe here