AI: Overview, Insights and Roadmap for Non-Profits

What is Artificial Intelligence?

Artificial Intelligence (AI) refers to the theory and development of computer systems that can perform tasks that typically require human intelligence, such as speech recognition, decision-making, and pattern identification.? AI encompasses a broad spectrum of capabilities, from mimicking human actions and thought processes to acting and thinking rationally.?

AI is rapidly evolving and has the potential to revolutionize the way organizations operate. ?AI can help organizations become more data-driven to make well-informed decisions based on accurate and real-time insights.? Modern generative AI is built on top of large language models (LLMs) which are specialized algorithms that have been trained on vast amounts of data to understand existing content and generate original content.? There are public LLMs that are based on the contents of the internet, but organizations can also train their own LLMs based on the corpus of intellectual property owned by an organization.?

It is important to note that there are different types of AI.? Weak AI is programmed to perform a single task, while strong AI possesses the full range of human capabilities, including talking, reasoning, and emoting. Weak AI can perform specific tasks faster and more accurately than humans, but it doesn’t possess the full range of human intellect. Examples of weak AI include chatbots like ChatGPT, Gemini, Claude and Copilot.? Strong AI is still in its early stages of development.? ?Strong AI refers to machines or programs with the mind of their own and which can think and accomplish complex tasks on their own without any human interference, while also having consciousness and self-awareness. There is much debate about whether it will ever be possible to create machines with true consciousness.

Within Weak AI there are several categories, including productivity / performance AI and generative AI.? Performance AI is focused on finding data faster, creating and displaying insights from data that may normally go unnoticed and improving process and worker performance.? Generative AI is a type of artificial intelligence that can learn from and mimic large amounts of data to create content such as text, images, music, videos, code, and more, based on inputs or prompts.

AI in its current forms is best used when teamed with people and not as a replacement of people. (The term co-pilot is often being used to describe this mode of using AI.) It's essential to understand that AI is not a wholesale replacement for humans; it's a tool that augments our abilities, enhancing productivity and innovation.? AI will allow for tasks to be completed with fewer staff and that gives organizations a chance to either redeploy staff to other areas or potentially shrink the size of their organization by becoming more efficient.

For AI to be adopted inside of organizations, there needs to be trust in the system output.? Objective outputs are easier to trust as they can be validated, but when the system output is subjective the output is more difficult to validate.? In general people will more easily trust the objective output.? ??Trust is a cornerstone of AI integration and to effectively leverage AI, transparency and trust are crucial.?? There are four key transparency practices that will help people build trust through transparency around data and AI:

• Tell people where the data comes from
• Open the black box of data categorization, analysis and computation – Help people understand what goes on inside of the system
• Match data visualizations with audience – Visualizations should be easy for people to interpret.
• Share openly when AI is being used to create work product.

In addition to trust, there are several challenges that organizations will face in leveraging AI.? AI bias refers to the phenomenon of artificial intelligence systems exhibiting prejudice or discrimination against certain individuals or groups. AI bias can occur when the data used to train an AI system is biased, or when the algorithms used to make decisions are biased.? Similarly, AI blind spots are unconscious biases present in organizational culture and data that organizations must ensure systems are architected around to uncover to avoid unintended consequences.? Generative AI has a problem referred to as hallucinations in which responses generated contain false or misleading information.? As organizations begin their AI journey, they must be committed to responsible scaling to be able to learn from its use before they continue to grow their uses and data sets.

Regulation and governance have not kept pace with artificial intelligence.? Regulation is trying to catch up, but still not able to keep pace.? Several important developments to point to are the AI White House Commitments made by the largest tech firms in AI, The White House and big tech companies release commitments on managing AI : NPR and the executive order issued by President Biden on October 30, 2023 for safe, secure, and trustworthy AI, FACT SHEET: President Biden Issues Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence | The White House.? Organizations must ensure they augment existing governance processes and create new where necessary to account for both the power and promise of AI, but also the potential negative consequences that accompany its use.

AI and your organization

Integrating AI into missional and business contexts

To be able to even begin thinking about the potential benefits that larger scale use of AI within your organization brings, you must examine your why when it comes to AI’s usefulness to the organization.? The first step to understanding and affirming your why is to conduct a readiness assessment that looks across the organization.? This assessment for the organization will help to begin thinking about how the organization can and should start to leverage more artificial intelligence.? There are many potential uses for AI at the organization and within its broader context, but the organization needs to be ready to embrace the idea of using AI, supporting its development and providing feedback during the various stages of development.

In this assessment, you will need to establish non-negotiables for lines that will not be crossed in deployment of AI capabilities.? This will need constant review based on your journey and is likely a moving line over time.? You must also develop ethical tenets that will be adhered to through the stages in the AI lifecycle:

• How do we build it?
• What do we put into it?
• What do we do with the output?
• How do we mitigate the unintended and inevitable consequences?

This assessment should help understand that weaving AI into the fabric of the mission and organization will not only help to enhance operational efficiency but also help to pioneer a new era of data-driven, impactful mission, dedication and philanthropy. Embracing AI isn't just about staying relevant—it's about leading the charge in a world that increasingly relies on informed, timely, and effective interventions.?

An assessment will also help to mitigate pitfalls that early adopting organizations are running into.

Inside of an organization you will have institutional use and individual use of AI.? For both types of use, you must factor in safety and security of the data that is being used with the AI.? Institutional use will be AI that you develop and control against data that we have inside the organization paired with external LLMs and data sets.? To ensure safety and security in this type of use, you must bring external LLMs to your data to reduce risk and improve the value to the organization versus sharing our data openly with the external LLM.? Good data underpins the institutional use of AI and we must remember that your data has built in biases.? You must also be sure that we believe in the quality, completeness and usefulness of your data to avoid mistruths from data interpretation.

Individual use of publicly available tools like ChatGPT, Gemini, Claude, Midjourney, Stable Diffusion and Copilot is already happening and will accelerate.? Staff need to be mindful of the data fed into these tools and the organization must invest in technology to prevent accidental leakage of sensitive data into them.? Staff in their use must also be mindful of what they are using in their work and whether or not it was generated by AI.? New watermarking and content credentialing tools are evolving but are not 100% in identifying human generated versus AI generated content.? Staff using non-standard software must be diligent in knowing that vendors maintain rights to use data that is input into their system.? Some vendors are even changing their terms of use to give them broader rights to any data going through their system for the purposes of their AI aspirations.

For both use cases you will need to augment existing policies and develop new governance and policies to protect the data, interests, and people that we serve.? These policies and governance need to factor in legal, ethical, theological, and structural factors to ensure we are comfortable with both the institutional and individual uses of AI for the organization.

Education is a critical piece to unlocking the value of AI and avoiding some of its pitfalls.? As you journey through the assessment, you need to in parallel start educating the organization on various aspects and understandings of AI and what it can mean for the organization.?

AI has the potential to help with the democratization of innovation for the organization.? When trying to get more innovative ideas flowing through an organization there are typically some challenges including:

• Evaluation overload – too many ideas
• Curse of expertise – More feasible ideas than novel ideas
• Lack of domain expertise – novel ideas that are not feasible
• Troubles seeing the forest through the trees

AI can help with overcoming these challenges by:

• Promoting divergent thinking
• Challenging expertise bias
• Assisting in idea evaluation
• Supporting idea refinement
• Facilitating broad collaboration with and amongst users

Use cases for artificial intelligence will abound for your organization as you begin to think and explore the opportunities for its usefulness.? Creative uses of AI and those that foster productivity are major areas that enterprises and organizations like ours are finding value in AI.? AI is not necessarily a replacement for staff, but in many cases an augmentation that will supercharge our creativity and productivity.? AI should instigate a change in your normal thinking.? Creative and productive / performance AI cuts across each of the areas I outline below.? Through the AI readiness assessment, you will be able to understand potential for these use cases as well as deduce new use cases, which will bring the most value and how to begin working on them.? A recent Microsoft survey said that 89% of those who had access to automation and AI-powered tools said they felt more fulfilled because they could spend time on work that really matters.? There are several ways to categorize AI use cases, but I chose one method to elicit conversation and exploration to conceptualize the most useful scenarios for AI.

Enhance Outreach & Engagement

• Fundraising & donor engagement
• Customer Service
• Communication
• Digital resource locators
• Go online to get offline - matchmaking, relationship making, community building

Optimize Resource Allocation

• Writing
• Financial Process automation
• Coding
• Internal knowledge searching and knowledge base creation
• Meetings - "Few things in life feel as wasteful as time spent in a meeting you didn’t need to attend."? Anonymous
• Worker enablement – A whole new way to think about work.?
• Customer Service

Amplify Impact

• Predictive analytics through machine learning
• Fundraising
• Programmatic applications

Foster Collaboration

Plan Forward

An organization has an opportunity and responsibility to understand and use Artificial Intelligence in its mission and operations.? You do not need to move at light speed, but you need to move.? This is a case where the organization cannot afford to do nothing.? You must seek to understand the benefits while being mindful of the potential issues with leveraging this technology.? This will require all parts of the organization to be involved and be open to exploring the opportunities afforded by AI and not leave it to IT to figure out.? Without collaboration on AI across the organization, we will certainly fail.? This is not a technology initiative, but an initiative to reshape the operations and mission of the organization for the better.

Below is an outlined summary of next steps that you should consider to begin your journey.? This outline is not linear and some of the activities can happen in parallel.

AI Readiness Assessment

Define values

• Know your why
• What are your non-negotiables

Confirm benefits and risks

Discover and understand needs / opportunities / strategy / vision and where AI is already being used in the organization

Intersection of data, people, process, and technology

• Need to accelerate data governance work for AI to be great.
• Work on data limitations - Organization, tagging, missing data, etc.
• Data quality improvements are a must

Use Case identification, approach and prioritization

AI governance Structures

Oversight & Monitoring

• Protect community
• Prevent bias
• Responsible scaling
• Legal, ethical, fair, and climate alignment

AI Operations

• Collaboration and feedback
• Conduct ethical impact assessments
• Ensure compliance with regulations
• Education & enablement

Education and Training

What is the good and the bad of using AI?

What are good prompts in using Generative AI?

What are AI usage guidelines?

Transparency internally and externally to where it is being used

Blind spot identification

Understanding AI capabilities in our contexts

Safety, security and privacy

Build a great understanding of data and intersection with AI in the organization

What do some simple examples of this technology look like for the organization?? Continue to run and demonstrate proofs of concept.

Policies

Principles of responsible AI to be added into existing policies or create new policies where needed: accountability, inclusiveness, reliability, safety, fairness, transparency, and privacy and security

Clear AI Guidelines and Polices - Build on existing

Use case execution - Start small and grow the ideas

Proof of concept

Pilot

Rollout

Monitoring / feedback loop with ROI metrics and analysis to iterate upon

Conclusion

In conclusion, AI is not just technology; it's a missional and operational tool to empower organizations in achieving their missions more effectively and efficiently. By understanding its capabilities, integrating it strategically, and addressing utilization concerns, AI becomes a faithful companion, enriching our work and connecting us with our community in profound and innovative ways.