Download the paper – What’s your endgame in the age of AI?

In 2015, Alice Gugelev and Andrew Stern published one of the more useful pieces of thinking the nonprofit sector has seen in recent years. “What’s your endgame?”, in the Stanford Social Innovation Review, made a simple observation: most nonprofits are asking the wrong question. By focusing on how to scale up, they pursue something the structure of the sector makes nearly impossible. Gugelev and Stern proposed six alternative framings – six “endgames” – that better reflect what most organisations can actually achieve.

I’ve been returning to that framework while researching how AI disrupts civil society. What I found is that the question is more urgent than ever, but the menu has changed. Some of the original six endgames have been strengthened by AI. Others have been hollowed out. Five new ones, at least, have emerged that didn’t need to exist in 2015.

The paper linked below is my 1st attempt to update the framework. It covers all 11 endgames – the six original, reinterpreted for the AI era, and five new – and includes a section on what this means for how organisations think about strategy.

This is a draft. I’d welcome feedback.

I’m sharing it now partly because two recent conversations have made me think the framework is incomplete in ways worth naming.

The 1st – a conversation with a colleague who leads an NGO working on labour conditions in Bangladesh’s garment sector – surfaced something that doesn’t fit neatly into any of the 11. His organisation exists not to produce expertise, but to generate ground truth that would otherwise not exist at all: to be present, observe, and make legible the subcontracting chains and informal labour arrangements that AI systems have no way of reaching. This is different from “domain expertise capture” (endgame 8), which assumes the knowledge exists and needs protecting. What he described is something that precedes this – the continuous act of making reality legible in the first place. I’m calling this potential endgame “ground truth generation”. I’m still working out whether it’s distinct enough to stand alone.

The 2nd concerns geography. The framework as it stands is written primarily for UK and US nonprofit contexts. Several endgames – government adoption, commercial adoption – don’t translate well to contexts where the state is structurally aligned with industrial interests, or where the commercial sector is itself the problem being addressed. That’s not an edge case. It applies to a large portion of civil society globally. And it exposes the inherent colonialised nature of the systems, structure, support and strategy that envelopes civil society, exhibited in the original 2015 paper and in my 1st attempt at an update.

I intend to do more research to make this framework relevant for what I think of as the “Global Whole” – not just the Global North nonprofit context it implicitly assumes. That research isn’t done yet. I’m flagging the limitation now, rather than waiting until it’s resolved.

If you have thoughts on either of those gaps, or on the framework more broadly, I’d welcome them. Please get in touch.

Download the paper – What’s your endgame in the age of AI?