AI in social value reporting — the GOOD, the BAD and the way forward

Everywhere you look, it seems like the whole working world is talking about Artificial Intelligence (AI). But one area in particular which poses some unique questions about the applications of AI is social value reporting.

Social value encompasses the economic, environmental and social benefits that individuals, organisations and governments generate to create a positive impact on communities and society. As businesses are increasingly held accountable for their contributions to social good, the application of AI in this space has opened up new possibilities for scaling impact, solving systemic challenges, and advancing equitable solutions.

But that doesn’t mean it’s without its challenges.

In this article, we’ll discuss how social value reporting professionals can leverage all the best bits about AI, without falling foul of the potential downsides.

AI in social value reporting: the GOOD

AI’s potential to do social GOOD can’t be overstated. A report from McKinsey & Company analysed 160 AI social impact use cases, and concluded that “existing capabilities could contribute to tackling cases across all 17 of the UN’s sustainable development goals.” In other words, AI has the potential to help hundreds of millions of people across the world, from all walks of life.

Plus, when it comes to the act of measuring social impact, AI can help you…

• Save valuable time and resources: At its core, AI thrives on data. Whether it’s about identifying trends, predicting behaviours, or optimising decisions, AI’s value lies in its ability to process and analyse enormous datasets. In the social value space, this capability is invaluable. Governments, investors and businesses looking to address societal issues are often flooded with data, but can lack the resources to fully utilise it.
  
• Spot hidden opportunities: Ask any social value lead about their biggest challenge, and they’ll likely point to the overwhelming volume of disparate data sets making it difficult to extract meaningful insights. However, advancements in predictive analytics have enabled AI systems to not only analyse this data but also identify and prioritise critical areas for intervention. AI can now uncover key themes, trends, and sentiments by detecting patterns in areas like unemployment rates, healthcare disparities and food insecurity. This allows decision-makers to focus resources where they are needed most, leading to more targeted and impactful actions.

• Reduce unconscious bias: No human is infallible. Despite our best efforts, we can all inadvertently allow personal biases to influence our decision-making — and that’s particularly bad news in the world of social value. But AI systems, if properly designed and trained, can evaluate data based on predefined criteria. Not only does this reduce the influence of any biases, it can also lead to more objective outcomes.
  
• Give more power to more people: It’s also worth noting that AI could be a vital tool in the democratisation of social value reporting. It offers smaller organisations (which may have previously lacked the required resources or skills) a more affordable means of investing in their social impact and scaling successful programs.
  
• Predict future challenges: Predictive models can be used to foresee potential societal risks, such as the spread of diseases or the rise of social inequalities. This forward-thinking approach enables organisations to proactively address challenges rather than reactively respond.

A deeper dive: improving access and inclusion through AI

A significant aspect of social value is the promotion of equity and inclusion. AI is uniquely positioned to address disparities by improving access to services and creating more inclusive systems. AI-powered tools can break down barriers for marginalised communities in several ways:

• Built environment: A local needs analysis can help identify opportunities in the areas of employment, crime, inequality, service gaps, housing, and air pollution. AI could be used to develop a targeted approach to maximise the social value produced in an area. 
  
• Healthcare: AI can enhance access to healthcare by diagnosing conditions remotely through telemedicine, making quality care available in underserved regions. AI-driven diagnostics and treatment planning are particularly beneficial in areas where healthcare providers are scarce, ensuring that more people can receive timely and accurate care. Some real life examples include helping people with low vision see the world around them with ChatGPT, diagnosing cancer from photographs, and cementing AI’s role within the growth of telemedicine.
  
• Education: Adaptive learning technologies driven by AI are revolutionising education, tailoring content to the unique needs of each student. This personalisation helps students who may not thrive in traditional learning environments, including those with disabilities, those in underfunded schools, and those from disadvantaged backgrounds. For example, the University of San Diego identified 43 examples of AI in education (ranging from classroom management to plagiarism detection), while teachers in Brazil are already reducing the time they spend on administrative tasks by leveraging AI.
  
• Financial inclusion: AI is helping to bridge the financial inclusion gap by providing alternative credit scoring models that assess individuals with little to no credit history. By analysing non-traditional data points such as mobile phone usage or utility payment history, AI enables underserved populations to access loans, mortgages, and financial services that were previously out of reach. Even better, it can bring fairness to lending services (reducing the biases faced by those applying for loans), while also increasing credit approvals — with no change in loss rates.

AI in social value reporting: the BAD

There are a number of distinct dilemmas when it comes to using AI in social value reporting. Let’s take a look at each of those problems, and what steps you can take to avoid them.

Problem #1: Taking people’s jobs

The problem: Ironically, while AI can be a useful tool for social value reporting, it has the potential to impact society in a much more obvious way: taking people’s jobs. Many fear that the efficiency of AI could see it replace humans in several critical roles. If managed incorrectly, AI could easily do more social bad than social good.

The solution: As is often the case, balance is essential. Our focus should be on developing AI tools that let social impact experts unlock new ways of measuring and interrogating data, empowering those individuals rather than making them redundant. It’s also worth mentioning that AI has the potential to create new jobs that don’t currently exist. Remember, the internet created more jobs than it destroyed, so there’s a potential digital inclusion and skills benefit for those looking to make a difference in their local communities.

Problem #2: Handling the data

The problem: Social value reporting can’t exist without data. And AI systems really, really love data. The key question is… how will AI access that data? How will it collect it, how will it store it and how will it use it? The privacy concerns are obvious.

The solution: Taking preventative measures now is key if we want AI and social value data to exist in harmony. Clear data ownership policies and ethical guidelines are an absolute must, as well as taking the appropriate steps to comply with any changes in GDPR regulations. Having a single source of truth in your business (like Impact’s powerful social value reporting software) is also invaluable, allowing you to combine all your data sources and build a more holistic story.

Problem #3: Not a real substitute

The problem: Will AI allow you to generate social value reports at speed? Yes. Are those reports a substitute for what industry professionals can do? Absolutely not. When AI generates results from a specific prompt or data set, it can be difficult to understand exactly how the system came to its conclusion.

The solution: AI is a useful starting point for building reports or frameworks, but human intervention is non-negotiable. That way, you can save time, without worrying about any poor decisions going unchecked. At Impact Reporting, we’re developing an AI feature that will make it easier to extrapolate your data from our system and build data-rich reports. It can empower you by saving time and uncovering new trends, but it only works with you in control.

Problem #4: Turning humans into data

The problem: The reason why ungoverned AI systems pose such a problem within social value reporting is due to the nature of what’s being measured. A simple mistake won’t just mess up a spreadsheet, it could genuinely lead to harmful consequences in the real world. AI systems simply can’t navigate the inevitable ethical grey areas of social value reporting, because they see the world around them as data — and nothing more.

The solution: Again, human intervention is paramount. Unlike an AI system, we can read between the lines, and see the real-life stories going on behind the data. AI can improve efficiency, but we should never forget that the decisions we make impact real lives, and not just a line of code.

Problem #5: Potential biases

The problem: In theory, AI is objective; it responds to the information it sees with logic, and its decisions are free of emotion or unconscious biases. However, if the data that an AI model is trained on contains inherent biases, the AI will treat those biases as gospel. They’ll worm their way into your social value reports, impacting outputs without you ever realising they’re there.

The solution: Ensure your data is as transparent and accurate as possible before handing it over to any AI systems. It should be up-to-date, comprehensive, and adequately reviewed beforehand for any potential biases it might reflect. Doing so will lead to more objective outcomes in your reports, and therefore better outcomes for the world. Outcomes that you can trust, too.

In summary…

There are many use cases for AI in social value reporting, but none of them are without their potential pitfalls. To leverage AI responsively, our focus should be on:

• Using robust models.
• Explaining and justifying AI outputs.
• Understanding how AI will use your data.
• Empowering humans rather than replacing them.
• Making AI accessible to all.
• Constantly improving each system.
• Ensuring everything is overseen by a human.

Did you know? Impact Reporting will soon be releasing AI-driven reporting features.

For more details about using AI safely in social value reporting and a sneak peek at what’s coming, contact Impact Reporting, or ​​get in touch with me directly at matt.haworth@impactreporting.co.uk.