On the difference between access and power ←
A note before the India–New Zealand AI summit; what I'm talking about tomorrow.
Tomorrow I am in Auckland to sit on a panel about AI, representation and trust. It is part of a summit between India and New Zealand, two countries that are not the centre of the AI world, deciding to have the conversation anyway.
I spent last night writing my answers. And somewhere in the middle of it, I caught myself about to write a word, and stopped.
The word was access.
The question I was answering was the big one. How do we make sure AI creates opportunity without widening the gaps that are already there. Access is the answer that comes to you first. Get the tools into more hands. Get more people trained. Open the door wider. It is a good answer. It is also the comfortable one.
Because here is what I have learned building this work across twenty-eight countries, most of them a long way from the places where these systems are designed. You can hand everyone the same tool and still widen the gap. All you have done is speed up the people who already had the data, the capital and the connection. Access on its own does not move power. It just makes the existing distribution faster.
The harder answer is the one I did not want to write at eleven at night, because it asks more of all of us. It is not access to the tools. It is power over the design. Who frames the problem in the first place. Who owns the data. Who gets to decide what success even means, and who gets to say a system is finished.
This is the part of AI I keep coming back to, and it is the part nobody can automate. The technology was never the hard bit. The hard bit is human. Who is in the room. Who the system was quietly built to work for. And who inherits the gaps without ever being asked.
Black and Brown communities, Māori, Pasifika and South Asian communities make up most of the people on this planet. Yet they are thin in the training data, thin on the design teams, thin in the rooms where the rules get written. When no one names that absence, it stops being a social gap and becomes a technical feature. The system performs worst for the people who were never in the room. Then we act surprised.
For a long time I thought the inclusion conversation was about adding people to something already built. I do not think that anymore. By the time a system is built, inclusion is just a review at the end. The real work is at the start, in the framing, when the questions are still open and the answers are not yet decided.
Which is why I find this particular pairing of countries quietly hopeful. India brings scale, remarkable talent, and real experience building digital infrastructure for more than a billion people. New Zealand brings something different. Trust, governance, and a hard fought conversation about Māori data sovereignty that most of the world is only beginning to have. And the more I sit with it, the more those two things look like the same fight. Communities deciding how their own data and knowledge are used, and who benefits when they are. Indigenous data sovereignty here. Protection from extraction everywhere. Same principle, different accent.
Neither of these countries is the dominant tech centre. That used to feel like a disadvantage. Now I think it might be the point. The standard for responsible, culturally intelligent AI does not have to be set by a handful of cities on the other side of the world. It can be set in rooms like the one I will be in tomorrow.
I do not know exactly how the panel will go. I rarely do. But I know the question I am bringing, and it is not really a technology question at all.
Not who gets access. Who gets to frame it.
If you have been sitting with the difference between being handed a tool and being given a say, I would love to know what it looks like from where you are.