Getting Started with GenAI

Intended audience: Social Impact Professionals

Social impact organization professionals who have not started to make use of GenAI or just started out testing different tools often come to a point where they need to make active decisions such as:

• What value does/can GenAI actually bring for us?
• What skills/knowledge and tools should we focus on?
• What are the ethical and legal risks of using GenAI? Should these prevent me from using GenAI for certain use cases? If not, how should I mitigate them?
• Where are our own guardrails? How can we follow through with them?

If these questions have come up in your work, then you’re in the right place. To start working on these questions, you need a dedicated team or person working on an overarching concept on how to use GenAI within the organization, as well as the trust and support of the management team or leading entity.

To summarize, here are 6 key steps to get started with implementing equitable and accountable Generative AI within your organizations for positive social impact:

1. Baseline learning: Assess your organization’s current knowledge of GenAI, data maturity, and technical literacy for different application approaches. This helps to get a clear picture of the level of knowledge, which tools are currently in use, and which are of potential relevance. Start with a simple assessment of the current status of GenAI usage at your organization. This could be done via a short survey, including assessments such as:
   1. Current knowledge of GenAI (possibly alongside traditional AI and data science)
   2. Current usage of GenAI (possibly alongside traditional AI or working with data in general)
   3. Potential and current use cases
   4. Tools and technologies of interest and currently used within your organization
   5. Already known blockers for working with GenAI
2. Define risks and learning paths: Use this assessment to determine the existing risks and need for necessary learning materials and training. Note that different groups within the organization will require a different kind of training (i.e. introductory AI lectures, prompting workshop, LLM hosting or fine-tuning, RAG Systems, ethics and privacy training, etc.)
3. Turn values to policy: Define and share guidelines to build your foundation with standard operating procedures and/or policy based on the assessment of potential ethical and legal risks. For example, here’s DataKind’s internal draft policy for GenAI use. Identify clear guardrails for your organization in alignment with your values. Formulate your organization’s value-aligned responsible AI principles, but also legal guidelines on using GenAI, which should be known to everyone within the organization.
4. Establish dedicated communication channels for exchange on GenAI: This could be an “Ask Me Anything,” AI debunking, or just use case discussion sessions; a slack channel; or an internal newsletter. What you select should depend on the size/structure of your organization, the importance of this topic, and your internal communication norms.
5. Evaluate opportunities: For all your possible use cases, first consider the challenges, constraints, and guardrails for risk mitigation. Prioritize relevant use cases to define a “vision” for GenAI at your organizations, using the next article in the section of the Playbook.
6. Set up tools and integrate them into your technology stack to give proper access to everyone. Guidelines for this step can be found in the article “Setting up GenAI Tools and Technology”

Getting started with software you already have

Regardless of the use case, specific tools that are currently available or might become available, many tech solutions you might already be using today have GenAI embedded in them that can be used for productivity improvements in a variety of ways. Common examples include:

• Zoom and other meeting platform producing meeting summaries or providing AI assistants to catch you up on conversations
• Salesforce navigation
• Google’s Bard features embedded in its products and Microsoft’s Bing features embedded in its products
• Many social media platforms have GenAI content creation features

Whatever software you use at your organization, look into it and you might find some GenAI features at least ready for testing.

Contributer(s): Matthias Boeck, with help from Rachel Wells, Deborshi Goswami, Daniel Nissani, Caroline Charrow

Contact us

If you would like to learn more about us, partner with us, or get in touch, visit our website or email community@datakind.org.

Subscribe to our newsletter
Subscribe