Empowering Humanitarians: AI, Data Integrity & Impact Reporting
As Idealists Rob Jordan and Cory Woolman head to NetHope and AidEx conferences next month, we’ve been thinking deeply about data, AI, and how these concepts play out in the world of humanitarian work. We hope you find the guidance below helpful in demystifying these developments in technology, and informative in planning adoption strategies.
Are you attending NetHope or AidEx this October? Find us at our booth at both events and catch our AidEx session: Empowering Humanitarians: AI, Data Integrity & Impact Reporting with Madeleine Fackler of IRC, Zak Kaufman of Vera Solutions, Dustin Pitts from Salesforce, and Gunjan Gupta of Twilio! Find more information about our upcoming events here. Read on for a sneak peek at our session, diving into how we’re thinking about AI, data integrity, and impact reporting for humanitarian organizations.
AI holds exciting potential with its ability to analyze, evaluate, and synthesize information. However, there are some fundamental steps nonprofits must consider before leveraging AI to maximize its capabilities. To make it simple, we have compiled the top three considerations for organizations looking to leverage AI.
Consideration 1: Maintain Data Integrity
Data integrity refers to the accuracy, consistency, and reliability of data throughout its lifecycle. It ensures that data remains unaltered and trustworthy from the moment it is created, processed, stored, and transmitted, until it is no longer needed. Maintaining data integrity involves implementing safeguards against prohibited modifications, errors, corruption, or unauthorized access that could compromise the quality and reliability of the information. Data integrity is crucial when concerning AI tools. If your data is unreliable, you risk potential AI hallucinations–and not the good kind.
AI hallucinating occurs when AI tools extrapolate a response from probability of the information that has been provided for analysis. In order to curb hallucinations you have to make sure your data is clean, true and organized. The cause of AI hallucinations is "data noise", which refers to irrelevant, redundant, or incorrect data that can disrupt the integrity and accuracy of a dataset. When AI assesses a situation using data with excessive data noise several challenges can arise, including:
- Reduced Accuracy: The presence of data noise can lead to incorrect conclusions and predictions, reducing the accuracy of AI tool outputs.
- Biased Results: Data noise can introduce biases that skew AI's understanding of a situation, leading to biased recommendations or decisions.
- Unreliable Insights: Excessive data noise might cause AI systems to generate unreliable insights, making it difficult to extract meaningful patterns or trends.
- Increased Computing Resources: Dealing with noisy data might require more resources to clean, process, and analyze the data effectively.
- Model Overfitting: Data noise can lead to overfitting, which occurs when AI models learn to replicate the unreliable data being provided rather than identifying actual trends or considering a larger perspective. This can cause poor generalization of new data points.
To mitigate these challenges, it's important to focus on data preprocessing, cleaning, and augmentation techniques that help reduce the impact of noise on AI model performance.
Consideration 2: AI Integration
Ask the Right Questions
Once your data is clean and ready for analysis, AI can play a significant role in analyzing data to produce impactful reports that showcase an organization's efforts and outcomes. In the interest of simplicity and technical accessibility, here are five ways AI can be used for this purpose:
- Natural Language Processing (NLP): AI-powered NLP can analyze textual data from various sources, such as customer feedback, social media, and reports, to extract insights about public perception, sentiment, and reactions to an organization's initiatives.
- Sentiment Analysis: AI can determine the sentiment expressed in text data, helping to gauge how stakeholders feel about the organization's mission and their impact.
- Data Mining: AI can sift through vast amounts of data to discover patterns, trends, and correlations that demonstrate the organization's impact on different aspects of society, the environment, or the economy.
- Predictive Analytics: By analyzing historical data, AI can make predictions about potential future impact, helping organizations plan more effectively and align their strategies with expected outcomes.
- Data Fusion: AI can combine data from various sources and formats to provide a comprehensive view of impact, creating a more holistic and accurate representation.
By leveraging these AI-driven techniques, organizations can use data to gain deeper insights into their impact, communicate their efforts more effectively, and drive positive change
It is important to note that this bridge between data and AI is still (for the most part) facilitated by humans. Clean data is utilized to produce reports and then that data is manually input into ChatGPT (commonly) for analysis. Eventually this will be integrated for the typical user–removing the need for human facilitation. That said, In order to be prepared for this eventuality you will want to choose a platform built with these considerations in mind. Salesforce is one such platform. Salesforce provides a CRM, Einstein (AI) Data.com (a data lake with advanced curating capabilities to ensure data integrity). This combination of solutions will soon empower users to make thoughtful decisions regarding their mission using a seamless process that allows the user to aggregate, clean, analyze, and report on data relevant to their organizational goals.
Consideration 3: Reporting on Impact
Tell the Story Numbers Can’t
Impact reporting involves assessing and communicating the effects, both positive and negative, that an organization, project, or initiative has on various stakeholders, the environment, and society at large. It goes beyond traditional financial reporting by focusing on the broader social, environmental, and economic outcomes of an organization's activities. Impact reporting typically includes data, metrics, and narratives that showcase the organization's efforts to create sustainable and meaningful change. This type of reporting helps staff, donors, and the public understand the organization's commitment to social and environmental responsibility.
Effectively delivering impact reports to your audience involves using report types that are clear, concise, and relevant to the organization's goals and mission. Here are examples of some of the more popular impact report types that can be used to share the analyses provided by AI from your (clean) datasets:
- Executive Summaries: Provide high-level overviews of the organization's key achievements, outcomes, and impact in a succinct format. These summaries are ideal for busy executives and decision-makers.
- Visual Dashboards: Create interactive dashboards with visualizations that display key performance indicators (KPIs) and impact metrics. Dashboards offer a quick and easily digestible way for staff to track progress.
- Infographics: Use visually engaging infographics to present data and statistics related to impact. Infographics simplify complex information and are well-suited for conveying key points.
- Progress Reports: Regularly update staff on ongoing initiatives and their progress toward achieving impact goals. These reports help maintain engagement and show the incremental changes being made.
- Comparative Reports: Compare impact data across different time periods, projects, or regions. These reports can show trends and improvements over time.
Ultimately, the choice of report types should align with the organization's culture, the preferences of its staff, and the nature of the impact being measured. A combination of different report types can ensure that the information is communicated effectively and resonates with staff members at various levels within the organization.
Effectively harnessing the power of data and AI to generate impactful reports involves a strategic journey comprising three key considerations: Data Integrity, AI Analysis, and Impact Reporting. The fusion of these three components creates a cycle of continuous improvement, enabling organizations to navigate the complexities of impact assessment and communicate their efforts transparently. This holistic approach empowers organizations to drive positive transformations, backed by data-driven insights and narratives that inspire action.
Our team of consultants is constantly working to stay up to date on the latest in technology innovation for humanitarian organizations. Sign up for our monthly newsletter to see what we're learning!