AI, Impact & the Double Standard of Automation

In the past months I've tried to get sort of a grip on the role AI can play in the world of impact entrepreneurship, and I have the amazing luck that I have quite some people flooding me with articles with the pros, cons and wows in it (I'm mainly looking at you Kurt Peleman , Lieve Vereycken , Ravi Bellardi and Astrid Leyssens !).

And one the things that right now really fascinates me are multi-agent models. Not just for the sake of the buzzword, but implications it has. The good, the bad, the ugly. So... Here's an attempt to bring together my reflections and open questions together in something somewhat coherent (I hope).

Relevance Over Novelty

AI innovation is often framed around novelty—what’s new, what’s disruptive, what’s next. But in the world of philanthropy and impact-driven initiatives, the default lens is different. There, the discussions on innovation aren't just about novelty, they're about relevance and impact.

That’s why I find it fascinating that philanthropy is investing so much in AI right now. Not because I work in that industry, but because it serves as a space where AI is evaluated not for how advanced it is, but for whether it actually makes a difference. For me, it's my go to sector to get a critical-curious voice on AI and it's (potential) role in Impact.

Multi-agentic models

Over the past few months, I’ve been exploring multi-agent AI systems — how AI agents don’t just analyze data, but interact with each other and a user, negotiate, and make decisions. And the potential of this in sustainability, impact, and knowledge-building is huge.

A few articles recently stood out to me, each offering a different perspective on where this is going:

• AI agents building a society in Minecraft: A company let 1,000 AI agents loose in a digital world, and they autonomously created social structures, economies, and governance systems. What happens when we apply this logic to real-world impact? Can multi-agent AI self-organize solutions for sustainability?
• The “Mirrorworld” and AI-driven simulations: This article explores how AI isn’t just predicting the future but creating digital twins of reality, where we can test impact strategies before deploying them. What if AI-powered policy simulations could prevent bad decisions before they happen?
• AI in philanthropy—automating decision-making: The Routledge Handbook explores how AI is being used to streamline funding decisions. But it raises an important question: If AI can automate aspects of social good, how do we look at this from a "keeping your job" pov? This is a reflection that I’ve been grappling with quite a bit, I'll rant about that a bit more below.
• AI-generated impact blueprints: A recent experiment used AI to generate entire strategic roadmaps for philanthropy. But does automation risk making solutions generic, losing the deep contextual intelligence that makes impact real? And if so, how can we counter this (which parts can be automated, which parts are deeply human?)?

I really believe in the power of a good interface between users and agents to make sure you're involved in the conversation (not just looking at it happening), and your feedback and opinions are fed back to a self-learning database / reference document in such a way that neurodiversity increases because you're interacting with the multi-agent model (whether or not we'll actually be able to make that happen is something we're still researching atm).

The Double Standard of AI & Jobs

One thing that stands out to me in every AI conversation: the panic over white-collar job automation.

There’s a hypocrisy in how we discuss AI replacing jobs. We’ve spent decades normalizing the automation of blue-collar labor—machines replacing factory workers, self-checkouts replacing cashiers, logistics AI replacing warehouse staff. But when AI threatens white-collar stability—writers, lawyers, consultants—the conversation suddenly shifts to panic.

Shouldn’t we be asking a different question? Instead of fighting to keep jobs unchanged, how do we rethink work itself? If AI can remove bottlenecks in sustainability and impact—translating knowledge faster, adapting solutions to context, making good decisions repeatable—shouldn’t we see that as an opportunity rather than a threat?

I see so many "learning networks" all over Europe to share insights and lessons learned in sustainability and circular projects. And very often these networks are highly ineffective in what they attempt. Not because of the idea of being a network. I love that. But because they focus mainly on transferring information and (often generic) learnings, and not on the actual networking.?

With LLMs and AI, it becomes much easier to ask the question "how can I make this about me". Sustainability is inherently context-dependent. By largely automating that part of the conversation, we can focus more on the hard part of it: "What does that imply?".

Some Open Questions

Right now I'm diving in head-first in learning the basics of automation with n8n (check it out, it's mind blowing how far low-code and no-code platforms have come!), I'm working further in exploring prompt-engineering, self-learning databases and how knowledge flows in innovation and impact, but most of all, I'm just trying to keep up with the bigger picture without getting lost in the tech development race AI is in right now.

So here are the 3 main questions or mindsets I have at the moment when looking at AI and automation:

1. Should AI agents be co-pilots in sustainability and impact, or just tools? (or put bluntly: Where are humans still involved? Sustainability is not always (rarely?) about how you feel about it.)
2. How do we use AI for real-world implementation, not just awareness? (Moving from insights to actions)
3. What does "work" mean in a world where AI can do more than we expected? (Could AI be an important part of the puzzle for moving from "doing more" to "doing better"?)

This is something I’m actively exploring with Manu Vollens and?ImpactGenie, but also?in our AI-powered design research project. If you're working on similar challenges—or just questioning where all of this is going—I’d love to connect.

?? Further reading:

• AI agents and emergent societies in Minecraft
• The Mirrorworld: AI and the Future of Digital Reality
• AI & Philanthropy – The Routledge Handbook
• Generative AI for Impact Blueprints