AI and Systems Change: Experiments in Augmentation

We all know AI can speed up work, boost efficiency, and cut down thinking time. Great. But let’s go further: AI can also supercharge our ability to tackle complex systemic change. It expands our access to knowledge, helps us uncover new insights, and augments our collaborative efforts — it’s about more than individual productivity.

ChatGPT Generated Image

Here’s how I think of AI as an augmentation tool in systems change:

1. AI does what humans can’t. Its ability to access and analyze vast amounts of information is invaluable in making sense of complex systems.
2. AI does what we can’t do given our constraints. It provides on-demand labor to do complex tasks — like offsetting 200+ hours of human analysis within a single month — when pressing questions need immediate answers.
3. AI does what we value, but not enough to prioritize. It can tackle tasks that are useful but not worth diverting scarce human resources for.

With this framing, here are some real-world examples of how AI is reshaping systems change efforts:

1. Initial systems sensing and systems mapping

AI platforms allow us to capture diverse perspectives and engage a wider audience in analyzing systemic problems and developing our baseline understanding of the system.

Check out Apurva’s systems sensing tools and network graphs. They source insights from stories shared across communities facing common systemic problems. This process broadens the knowledge base contributing to systems understanding while democratizing access to sense-making tools. I’ve been looking for an opportunity to tap into this amazing new platform.

2. Strategy design using a systems dynamics approach: Finding places of leverage

Referring back to a piece I wrote last year with Julia Coffman, many systems change strategies use a systems dynamics approach: finding a clear point of leverage, something worth changing in the system because of how it will ripple through the system.

This type of strategy design requires having a fairly high level of confidence in how change will happen. It relies on evidence about how change has happened in the system in the past, about how the system functions today, and about how changes in similar (though never identical) contexts have unfolded.

AI can supplement our own thinking by offering:

• Evidence from past changes in similar systems;
• Competing perspectives on which leverage points hold the highest value; and
• Competing perspectives on potential interventions into the leverage point.

3. Strategy design using an emergence approach: Finding experiments worth trying

Other strategies use a systems emergence approach, seeking areas within the system where smaller experiments are worth trying in order to see what happens. In these cases, we don’t need overwhelming confidence about whether this is the right place to intervene. Instead, we need to have ways of steadily seeing how the system is unfolding and how and why change is happening.

In emergence-based approaches, AI can assist by:

• Collecting real-time data and helping to analyze it in order to find patterns in how change is happening (e.g., check out the AI interviewer that is part of the Causal Map application);
• Rapidly analyzing patterns to generate action insights; and
• Helping organize insights that are coming in from many places, through visualizations or narratives.

I recently did this as part of a three-country strategy focused on racial equity in the education system. I’ll admit we had some hiccups (and I’ll get around to writing another blog about that specific process soon), but I also know the depth of systemic insight we were able to glean in the space of time we had available was impressive and not previously possible.

4. Ongoing sense-making

Ongoing systems sensing sessions are a critical part of responding to dynamic, complex systems. Here, I’ll define these as moments where the stakeholders involved have opportunities to talk about what they observe in the system, what is changing, why it’s changing, and how.

AI can play many roles, including:

• Acting as a key stakeholder during discussions, challenging our assumptions and offering systemic insights.
• Sourcing external insights and practical information in real-time.
• Sourcing external insights from stakeholders and organizing them to use during discussions.

I recently engaged AI as a “stakeholder” at the table during a systems sensing session, feeding it information about the strategy and system in advance, sharing with it some of our key insights as we talked, and ultimately asking it questions to help us challenge some of our assumptions. During this discussion, the AI reframed one topic we were exploring from focusing on individual behaviors to understanding systemic constraints and motivations — a critical shift in perspective.

AI can also be used to steadily source new insights from outside the room. In addition to Apurva, mentioned above, other platforms like Amoofy are able to help with this too. Amoofy was originally designed to bring value to individuals, families and communities by creating a way to share and curate their legacy of stories. Now it can also be used to curate stories across a larger context such as a systems change effort. Through collective storytelling and harvesting we can make visible how systems are changing and why.

5. Retrospective storytelling about the long arc of change

One of the things we don’t do enough is look back at the history of how systems have been changing. When we do, we often limit ourselves in time, geography, or sector or only focus on an evaluation of specific interventions.

AI can help us overcome these limitations and make visible the long arc of how change has really happened. I’ve been using AI for this purpose in a 10 year story of a wickedly complex system (related to slavery and forced labor in a global supply chain). It is a story that has both a country-level focus and a global focus; covers public, private, and non-profit sector changes; includes multiple policy arenas and issue areas; and ultimately had significant impact on the workers, documented through population-level studies.

AI has augmented the skills and capacity of our research team, including:

• Sourcing insights from thousands of pages of documents
• Helping us decide on which documents we want to manually code and analyze.
• Helping find patterns across the documents and insights.
• Helping find causal connections referenced in these documents.
• Helping to articulate key themes (causal chains across a set of interrelated parts of the system) to explore with stakeholders in group sense-making sessions.

What’s Next?

AI is opening new pathways to amplify human efforts in systems change. From sensing and mapping to strategy design and storytelling, it’s a powerful partner for tackling complexity.

Are you experimenting with AI in systems change work? I’d love to hear your examples!