Two Weeks in GenevaCarrying AI for Every Mind to the United Nations

I have stood in front of a lot of rooms. A conference hall in Bengaluru, a lecture theatre at Golden Gate University, a video call with a teacher in a district I'd never heard of before that morning. Nothing quite prepares you for the particular quiet of a UN meeting room before it fills — the microphones being tested one by one, the interpreters' booths glowing at the back, a nameplate with your name on it in a font that isn't yours to choose.

The week opened with a side event I moderated inside the UN Global Dialogue on AI Governance — co-organised by the World Digital Technology Academy, Golden Gate University, and ICARUS AI Inc., opened by Ambassador Prof. Muhammadou M. O. Kah, Vice Chair of the UN Commission on Science and Technology for Development. My task was simple to say and harder to do: keep four institutions, four instincts, and one room full of very different definitions of “access” moving toward the same point.

The session argued a single progression, and asked the panel to defend each step of it: AI Access → AI Literacy → Human Agency → Responsible AI Governance. Education isn't one application of AI policy — it's the foundation the policy has to stand on. I'd made that argument in classrooms before. I had never had to make it with the UN flag behind me.

“For the first time, AI answered me instead of someone answering on my behalf.”A visually impaired student, recounted during the roundtable on AI access

That line — from a classroom in India, long before it reached Geneva — became the thread the whole panel returned to. Access was never about whether someone could open an app. It was about whether they could question what it told them. When we turned to who is really “the human in the loop,” the room agreed on something uncomfortable: a teacher managing forty students, or a frontline worker who received a two-hour AI orientation, isn't really in the loop if they can't challenge what the system tells them. We've too often mistaken accountability for governance.

By the time the closing commitments went around the table, I realised the question I'd carried into that room wasn't really about AI at all. It was the same one my mother asked in her own classroom, decades before either of us had heard the word “algorithm”: does every student in this room actually understand what's in front of them, or only the ones we designed it for?

The last engagement moved from classrooms to something more foundational. Room Concordia 1 at UNOG doesn't look like a place built for tenderness — travertine floors, a banner instead of a stage, security thirty minutes deep. And yet the question sitting at the centre of it was the most human one asked all week: as AI reshapes the world, who shapes the values that guide it?

I grew up in a family where the water, the mine shaft, the classroom, and the school office each asked the same thing of the person inside them: serve something larger than yourself. My mother stood in front of a classroom and gave children the language to understand their world. My grandmother led a school as its principal. Neither of them ever used the word “governance.” Both of them practised it daily, in the only way it has ever actually worked — one person, choosing to be accountable for another.

The position I carried into that room was simple, and I said it plainly: empathy and responsibility aren't soft add-ons to AI safety — they're the compass, because only people can love, protect, and take responsibility for one another. Algorithms can optimise. They cannot care. If AGI is coming whether we're ready or not, the question was never whether to build it safely. It was whose values get to decide what “safe” means.

Alongside H.E. Amb. Muhammadou M.O. Kah, Dr. Mehdi Snene, Professor Yale Li, Ricquelle Jeffrey, and Professor Salma Abbasi — this was the panel where AI for Every Mind's premise met AI governance's biggest open question, head-on. It was also, quietly, the panel that felt least like work.