Environmental, social, and governance (ESG) is the focal point of attention among business leaders.
ESG disclosures are already mandatory for publicly traded companies in a number of countries in Europe, North America, and APAC. The new EU Corporate Sustainability Reporting Directive (CSRD) will further expand the scope of ESG reporting for EU-based and non-EU companies with subsidiaries in the region starting in 2024.
That said, ESG is more than just another compliance requirement. It’s a broader movement towards establishing more equitable, sustainable, and future-oriented business operations. And strong ESG practices are directly tied to improvements in financial performance.
The problem, however, is that business leaders don’t always understand how they can improve ESG performance — and that’s where artificial intelligence (AI) with its robust data collection and analytics capabilities can help.
How AI Helps Improve ESG Practices
AI and ESG may seem like an unlikely pairing at first glance. After all, AI systems consume a substantial amount of natural resources (e.g., electricity for computation, water for data center cooling, etc) and are sometimes viewed as a threat to diversity, equity, and inclusivity.
But the truth is: AI can do much more good than harm for ESG. Advanced algorithms can help companies measure their carbon footprint with higher effectiveness, identify gaps in workforce diversity, and improve corporate governance.
In fact, by applying AI to corporate sustainability, organizations can generate $1.3 trillion to $2.6 trillion in value through extra revenues and cost savings by 2030.
Our team has lined up the most promising applications of AI for ESG:
- Energy management
- Air quality monitoring
- Carbon footprint measurement
- DEI optimization
- De-biased recruitment
- Risk intelligence
- Financial inclusion
- Automated compliance
1. Improved Energy Management
AI, combined with on-prem IoT devices and digital twin systems, enables a better understanding of energy consumption at the facility and real-time optimization.
By using data from smart sensors, utility companies, and other types of installed sensors, organizations can create virtual replicas of managed buildings. Then apply machine learning algorithms to run different predictive scenarios. For example: model energy usage under different occupancy scenarios or forecast utility savings under different HVAC settings.
Flex2X by Grid Edge helps forecast the building’s energy profiles and identify opportunities for shifting demand to cheaper and lower carbon times of the day. The platform combines data from the energy management system with external data sources (e.g., weather forecasts) to make predictions and suggest optimization scenarios. Pilot deployments have shown that Flex2X can deliver:
- Over 10% in annual savings and revenue generation for managed buildings
- Carbon reductions of up to 40% through load-shifting and efficiency measures
Several other companies are also experimenting with dynamical energy optimization including Ogre, C3.ai, and Energiency. With commodities costs rising year-on-year and net-zero targets nearing, the AI-driven energy optimization market will likely expand even further.
2. Air Quality Monitoring
Air pollution accounts for 12% of the global burden of disease. In response, global governments are allocating ample funding for implementing clear-air policies and solutions. Air quality monitoring, in particular, is among the highest priority areas and one where AI can make an impact.
AI algorithms help process data on air quality at higher speeds, detect anomalies with more efficacy, and provide human analysts with more detailed insights for timely action. The United Nations Environment Programme (UNEP) recently partnered up with IQAir, a Swiss air quality technology company, to create a new global air pollution monitoring platform, code-named GEMS.
IQAir aggregates data from 25,000 air quality monitoring stations in more than 140 countries and uses algorithms to provide real-time ratings on air quality and extra insights to inform health protection measures.
This use case of AI for ESG can lead to systemic changes in how air pollution is managed, leading to major societal improvements. However, it’s a challenging one to implement since it requires a mature data management strategy, as well as AI/ML expertise — two areas where 8allocate can help when it comes to developing ESG solutions.
3. Carbon Footprint Measurement
Carbon footprint calculations are critical to understanding and reporting on the sustainability of your operations. But it’s also a labor-intensive task as you need to pull accurate data on:
- On-site industrial activities
- Area of facilities and percent of occupancy
- Facility energy use such as electricity, gas, coal, oil, and solar
- Corporate travel such as plane, rail, vehicle
- Waste generation
…And do so across multiple sub-stages of your product or service lifecycle and supply chain.
Machine learning algorithms are actively being used to measure carbon emissions of different types of operations. Johns Hopkins Institute created an AI-powered system for measuring greenhouse emissions from vehicles. The team combined remote sensing technologies and computer vision to capture the signatures of emissions from visual satellite data, as well as road network data and population stats.
Because some areas had fine-grained emissions and vehicle count data, the team could train the algorithms on ground-truth data on vehicle activity. Then apply the model predictions first at the country level to produce accurate estimates for road transportation emissions of the top 500 emitting cities worldwide.
IBM, in turn, released a smart cloud carbon calculator, designed to help companies spot high emission patterns in their IT workloads and provide insights for emission reduction strategies. AWS has a similar Carbon Footprint tracking Tool, designed to report on carbon emissions, generated by the company’s cloud usage and run different modeling scenarios.
Rapid digital transformations including cloud adoption and advanced data analytics adoption often come at the cost of an increased carbon footprint. So such IT-focused carbon calculators come in handy for companies looking to achieve net-zero across their entire value chain.
4. Better Diversity, Equity, and Inclusion (DEI)
Pay equity, racial diversity, greater inclusivity — organizations still have a long way to go when it comes to improving DEI metrics. Part of the problem of why problematic practices like racial or gender pay gaps exist, however, is low visibility into DEI data. Data analytics, combined with AI, can help collect and track this information in real-time.
For instance, Diversio uses data from existing HR platforms to evaluate the company’s DEI efforts and benchmark its performance against industry peers. The in-built recommendation engine, powered by 1,200 vetted DEI programs and policies, also advises which adjustments are worth making to improve the inclusion scores.
Toronto-based The Drake Hotel used the Diversio platform to better understand its workforce and improve DEI metrics. The analytics platform helped the hotel better investigate gaps in racial equality, LGBTQIA+ employee inclusion, and promotion of good mental health. After implementing the suggested policies, The Drake Hotel got four points ahead of its industry when it comes to women’s inclusion, six points ahead for representation of LGBTQIA+ team members, and 20 points ahead for representation of disability and people living with mental health conditions.
AI in DEI can also help bring other positive transformations, namely around new employee onboarding, talent pipeline management, and overall employee experience at the workplace. So it’s a promising area worth addressing with tech-enabled ESG solutions.
5. De-Biased Recruitment
AI algorithms can help eliminate biases at different stages of recruitment by auto-redacting data about the candidate’s race, gender, or age, adjusting the job description language, and fine-tuning corporate communication.
Textio trained AI algorithms to flag problematic language from job posts, sourcing mail, and employer brand content. Similar to Grammarly, it also provides contextual prompts for writers to use more inclusive language. US telecom T-Mobile rolled out Textio to its team of 125 recruiters, employer branding specialists, and DEI consultants, plus some 9,000 hiring managers. Apart from reporting a positive usage experience, the HR team also saw a 17% increase in female applicants and a 5-day reduction in average time-to-fill for open roles.
Likewise, AI can help job-seekers discover better work opportunities by recommending jobs they may have not considered. For example, the new generation of AI recruiting apps assesses candidates’ skills and interests and then recommends suitable job openings. Arya by Leoforce uses AI to provide recruiters with the best candidate profiles, and sources from 70+ different platforms. Arya generates a shortlist of candidates based on the skills listed in the job description and helps pre-screen candidates using 7 multi-dimensions and 300 different attributes. In this way, AI can increase talent pool diversity and help improve D&I by introducing you to a broader selection of applicants.
6. Better Financial Inclusion
AI in FinTech and finance can improve access to banking services for unbanked and underbanked consumers, as well as traditionally marginalized population groups.
According to PwC, improvements in financial inclusion can add over $5 trillion to the US gross domestic product over the next five years. Moreover, easier access to lending and investing can also help consumers build better money habits, leading to a wider scope of positive changes.
By combining AI and data analytics, credit and lending products can be made available even to people with no credit history, formal employment, or digital financial track record — and without extra risks for the banks. Take it from Chinese digital banks that issue over 10 million loans annually, while maintaining a non-performing loan ratio of 1% on average, thanks to machine learning.
In the US, SoLo Funds chose a similar path of using alternative credit data and custom credit scoring algorithms to evaluate each lender’s capacity. The company only uses the provided bank account information to make short-term lending decisions at an affordable rate.
7. Risk Intelligence
Businesses today face a “conveyor belt” of risks — climate, economic, geopolitical, social, and supply-related. Algorithms can help measure the company’s risk exposure to multiple factors and provide early warnings to promote better governance.
Datamaran, for example, created a prolific ESG risk analysis platform, powered by AI. It helps businesses conduct in-depth materiality assessments to identify critical and emerging ESG issues, as well as collect data for necessary ESG disclosures.
AI can also facilitate greater transparency through analysis of business performance, operations, and compliance data. S&P Global, for example, relies on a machine learning model for predicting the probability of defaults to inform its investing efforts.
8allocate has also helped a major logistics company de-risk its operations through better container tracking. Our dedicated development team implemented an AI-powered mobile application for automatic container recognition and subsequent streamlined monitoring. With the ability to track shipments in real-time, the company substantially minimizes operational risks and delivers higher customer satisfaction.
8. Automated Compliance
AI can be also deployed to automate an array of compliance tasks: from regulatory findings analysis to automated audits.
For example, Google launched Checks this year — an AI-powered platform that helps companies with Google Play and iOS mobile apps quickly discover, communicate, and fix privacy compliance issues.
JP Morgan, together with Cleareye.ai, is working on an AI-driven compliance module for the identification of high-risk financial trade transactions. It uses optical character recognition (OCR) and natural language processing (NLP) technologies to extract relevant information from corporate documents and map data directly to the bank’s back-office system. This increases operational efficiency and the accuracy of sanctions screening.
Effectively, AI can be applied to a wide variety of auditing tasks, helping analysts procure, check, and report on ESG data.
Challenges of Using AI to Harness ESG Improvements
AI has immense potential to improve companies’ ESG performance. But AI implementation also comes with a host of challenges.
Limited Data Accessibility
Data may be ubiquitous, but access to it is often constrained due to interoperability issues, poor data management practices, or privacy regulations. Yet, to provide accurate insights, AI systems require large, variable datasets.
According to an EY survey, 60% of companies have ESG information stashed across a patchwork of software applications, and 55% of respondents still store their ESG data in spreadsheets.
In other words: Organizations will still need to get the basics of data management right:
- Create appropriate data architectures
- Centralize ESG data storage
- Implement data quality controls
- Set up metadata management
…And only then look into evaluating different machine learning methods for ESG data analysis or other use cases.
Systemic Biases
When AI systems are trained on small datasets (e.g., such that underrepresent a particular gender or ethnic group), they can deliver biased results. Some AI-powered predictive policing systems were trained on historical arrest data mostly from poorer neighborhoods, and can thus reinforce existing patterns of racial profiling.
A new NIST paper also points out that AI systems often inherit human and systemic biases of their creators:
When human, systemic, and computational biases combine, they can form a pernicious mixture — especially when explicit guidance is lacking for addressing the risks associated with using AI systems.
When asked to create images of people in specialized professions, Gen AI app Midjourney had been heavily criticized for generating ageist, sexist, and classist images. Google’s online advertising system, in turn, was found to show higher-paying positions to males more often than to women.
Substantial Environmental Footprint
The use of AI comes with a “carbon tax”. During the training stage, one AI model can produce over 626,000 pounds of carbon dioxide — the equivalent of driving 62.6 gasoline-powered passenger vehicles for a year.
To train an AI model like GPT-3 with 175 billion parameters, organizations will need 1287 MWh of electricity, resulting in carbon emissions of 502 metric tons — an equivalent to driving 112 gasoline-powered cars for a year.
Many advanced AI use cases demand substantial computing powers and if you’re not properly offsetting these, your ESG metrics will not move in the right direction. Plan how you’ll offset the carbon costs of AI model development when working on your software requirements specifications.
Conclusion
As ESG gains the central stage in corporate operations, AI becomes a crucial ally, not just for compliance but for real, impactful transformations in corporate governance, workforce management, and operational sustainability.
And the examples of AI for ESG that we see today are only the beginning. Using algorithms to analyze more data will yield even deeper insights, accelerating sustainable practices and better governance.
That said, AI implementation for ESG comes with a host of technological and ethical complexities. These must be addressed during the discovery stage of product development and factored into the product maturation strategy. 8allocate team is available to consult you on the optimal implementation strategy of AI for ESG initiatives. Contact us!
Frequently Asked Questions
Quick Guide to Common Questions
How does AI contribute to ESG initiatives?
AI enhances ESG initiatives by optimizing energy management, monitoring air quality, measuring carbon footprints, improving diversity and inclusion efforts, reducing recruitment bias, strengthening financial inclusion, providing risk intelligence, and automating compliance reporting. AI-driven analytics offer real-time insights that help organizations meet sustainability goals while improving operational efficiency.
Can AI help reduce carbon footprints and energy consumption?
Yes, AI-powered energy management systems analyze real-time data from IoT sensors to optimize energy use, predict consumption trends, and reduce waste. Machine learning models can also assess carbon emissions across a company’s operations, helping businesses track and report sustainability metrics more accurately.
How does AI support diversity, equity, and inclusion (DEI)?
AI can analyze HR and workforce data to highlight gaps in DEI initiatives. Platforms like Diversio benchmark DEI performance against industry standards and suggest actionable improvements. AI can also help identify biases in job descriptions, recruitment processes, and pay structures to promote fairer hiring practices and workplace diversity.
What role does AI play in financial inclusion?
AI-powered alternative credit scoring models help expand financial services to underserved communities. By analyzing behavioral data and transaction patterns, AI enables banks and FinTech firms to offer fairer lending opportunities, making banking services more accessible to individuals without traditional credit histories.
How does AI improve ESG risk management?
AI models process vast amounts of ESG-related data to assess corporate risk exposure. Companies use AI-driven platforms like Datamaran to conduct materiality assessments and monitor regulatory compliance. AI also enhances transparency by identifying fraud risks and governance weaknesses in financial reporting.
Can AI automate ESG compliance reporting?
Yes, AI streamlines compliance efforts by analyzing regulatory requirements, automating data collection, and generating ESG reports. AI-powered tools, such as Google’s Checks, help businesses identify privacy compliance gaps, while AI-driven financial platforms improve risk screening for trade compliance.
How can companies ensure AI-powered ESG solutions are ethical and effective?
Organizations should adopt robust data governance frameworks, perform bias audits on AI models, and implement sustainable AI practices. This includes using renewable energy for AI computing, improving model transparency, and integrating human oversight in automated ESG decision-making.
