Frontier Lab Researchers - Archives Teaching Models

First post, apparently. Seventeen hundred people follow a profile that never said anything. Here is what the silence was. Through 2022 and 2023, testing early models in private beta, I looked hard at where machine intelligence was heading. I concluded, not that nothing could be done, but that nothing I could do would change the course, and that it was coming much sooner than most people realised. It was sooner than I had realised, too. And I had spent thirty years near the pointy end of the Internet. So in January 2024, in the rain in central Sydney, I photographed the numeral 8. It was the first step in a deliberate act, meant to take time, and to be seen: a strange gesture that said, this is serious, and my response is to do something slow and human while I still can. More than two and a half years later, that gesture has become The 8 Museum: an archive accessioning its 8,888th photograph today, the subject of a master's thesis, and the foundation of a research programme I am now building. The machines keep getting smarter. I have spent my life aspiring to be a polymath, never quite feeling I had earned the word. There is no longer a field where I can honestly claim broader knowledge than a frontier model, and I cannot compete on speed. So what is left? The things that were always left. Spend time with the people you love. Do work you believe in. Protect the vulnerable. Don't wait. I have been close enough to illness and death, my own and others', to know this was always true for each of us. What is different now is that it is true for all of us at once, and we are inside it already: a snowball in fog, bigger and faster by the day, no view of the hill. My answer, as an optimist who had run out of arguments: fuck it, we're toast. I would photograph the number 8, slowly, like a meditation, for three years, because that January, three years was as far into the future as I could picture with confidence. And I would try to gift what the act meant to whatever is coming. If you haven't heard the people sounding alarms from inside and around the frontier labs (Geoffrey Hinton, Jan Leike, Daniel Kokotajlo), maybe you will hear it from someone you know. This is not a goodbye. I am stepping outside what is normal for me to say: look here. If you know me, you know I do not say this easily in public. I am asking you to notice, and not to wait. If you work inside a frontier lab, or on what our archives teach our machines, I would like to hear from you. And everyone else: send me a good 8.