Hey friends,
I'm writing this from Geneva, representing Australia at the United Nations' AI for Good Summit.
I spent a lot of this week in rooms I didn't expect to change my mind. Governance rooms. The same week, Geneva hosted the UN's first ever Global Dialogue on AI Governance, every government at one table. I sat with people from the WHO, UNICEF, government delegations, some of the very people whose job is to write the rules for AI safety and security. I walked in expecting caution. Brakes. "Let's slow down until we're sure."
That is not what I heard.
The conversation was serious about risk. Deepfakes, child safety, testing, accountability, all of it on the table. But not once was the goal to put AI on hold until someone could guarantee it was safe. The goal was the opposite. How do we let progress and safety move together, so that safety strengthens the work instead of stalling it.
Nobody Was Waiting For A Definitive Answer
Here's the thing that hit me hardest, and it's the whole newsletter really.
None of these organisations had neat, final answers. And they weren't pretending to. Even the UN's own panel of top AI scientists publishes an evidence report, not a rulebook of certainties. They aren't claiming to know for sure. They're gathering the best evidence available and moving with it.
That's because AI is probabilistic by its nature. It deals in likelihoods, not guarantees. So the frontier has quietly accepted something most of us back home haven't. A definitive answer is never coming. Instead of waiting for certainty, they work to shift the probability in their favour. Test, experiment, learn, adjust, stay safe, go again.
Then I think about the conversations I have at home.
We Want A Guarantee Before We’ll Move
When I talk to a lot of Australian businesses and government bodies, the first question is almost always the same. Can you guarantee it will work. Can you guarantee it's safe. Can you promise it won't get anything wrong.
It's a fair instinct. But it's the wrong question for this technology, because the honest answer is no, not with certainty, nobody can. And so we do the most natural thing in the world. We wait. We form another committee. We ask for another report. We stay exactly where we are, using AI to write the odd email, and we tell ourselves we're being responsible.
We're not being more careful than Geneva. The people at the frontier take safety more seriously than we do, not less. The difference is they refuse to let the absence of certainty become a reason to stand still.
The Real Gap Is Permission
That's the gap. It isn't the tools, we all have the same ones. It isn't talent. It isn't even governance maturity, because the countries writing the actual governance are the ones moving fastest.
The gap is permission.
The organisations moving quickly gave themselves permission to experiment, to be wrong in small, safe ways, and to keep going. Nobody handed them that permission. They took it. Back home, we're still waiting for someone, a regulator, a vendor, a case study, anyone, to tell us it's okay to start.
Nobody is coming to tell you it's okay.
A Calm Takeaway
The frontier isn't ahead because it's more sure than we are. It's ahead because it made peace with not being sure, and moved anyway, carefully.
Certainty was never going to arrive. That was never the deal with this technology. Permission is the one thing you can actually give yourself, today, from your own desk.
So this week, change the question. Stop asking whether AI is certain and safe enough to begin. Ask instead where you could afford to be wrong safely, and start there.
That isn't recklessness. In Geneva, that's just what the grown-ups are doing.
See you next week,
— Aamir
📲 Resources & Links
🎧 Listen to the Podcast Episode on: Spotify | Apple Podcasts | YouTube
📘 Book: The CEO Who Mocked AI (Until It Made Him Millions) by Aamir Qutub