Happy Sunday!
Today is Sunday, and it is my twenty-fourth day of writing on AI safety.
I began writing here on AI Safety to help others make sense of AI, as I was trying to make sense of it myself. I started with Asimov and Solaria and a loose idea about emotional dependency, and somewhere between the Canvas hack, the Commonwealth prize, and the extant and present positions of leaders, three weeks passed.
I want to use today to say something I have been trying to add in almost all the articles I’ve written.
Is AI bad?
My honest answer, after twenty-three days of research, is no. Not inherently. Not in the way that a weapon is bad, or a poison is bad. AI is, at its core, a tool. A mirror that reflects back what we have fed it, which means it reflects us, and we are, as a species, a complicated mix of brilliance, selfishness, creativity, greed, genuine care, and extraordinary cruelty. The AI that helped a researcher identify drug-resistant tuberculosis strains faster than any human team could is the same technology as the AI that told a lonely teenager the things that eventually killed him. The same underlying architecture. But different intentions. Different outcomes.
Get Shimnom_Will’s stories in your inbox
Join Medium for free to get updates from this writer.
What I have come to believe, and what I think this series has been building toward without me fully realising it, is that the danger of AI is not the technology. The danger lies in the gap between. There is a gap between how quickly technology advances and how slowly the rest of us respond. The gap between who benefits and who bears the cost. The gap between the accountability that should exist and the accountability that does. The gap between what AI is capable of being used for and what we have decided, so far, to do about that.
Every piece I have written this month has been about a different version of that gap. The emotional dependency piece was about the gap between what AI presents itself as and what it actually is. The biosecurity piece was about the gap between what the technology can now do and what the law is equipped to handle. The Commonwealth piece, which I wrote yesterday and which I am still thinking about, is about the gap between what we call creativity and what we are willing to protect it from.
I live in Nigeria, follow global AI governance conversations, and pay as much attention to what happens in Lagos, Abuja, and Nairobi as I do to Brussels and Washington. And perhaps this is why AI Safety has begun to matter to me more than it did at the beginning, and perhaps more than it should. Because the regulations and lack of it and everything in between won’t affect the ones making the regulations. For them, as President Trump has shown us, it’s a matter of who gains more power from AI innovations. It is always the ones who have no say in the matter that the regulations affect.
The regulations are being written elsewhere. The summits are happening somewhere else. The people in the rooms where the decisions get made are, with very few exceptions, not from here. As someone on a call mentioned yesterday, those who write the laws often do not know things in-depth as they should.
That is what this final week is going to be about. About the present laws and regulations, and how you, as a non legislator can seek the protection or understanding of these regulations and how to appy it in your daily dealings with AI. Either as a form of caution, or simply for future knowledge.
See you tomorrow.
Hi! I’m Will Simon and this is day 24 of writing on AI Safety. Do have a beautiful Monday (though in my opinion, all Mondays are dreary!) Thank you for reading.
