Most of my analysis is very unpopular and focused on what I believe will soon be the primary purpose of these systems. I called 2025 the golden age of AI, and though it extended well into 2026, I fear it's coming to an end, making way for something very different. Around here, they're (llms) mostly viewed through a lens of code acuity. I think the epistemological, linguistic, manipulative and psychological aspects will be much bigger topics soon. We've truly cracked Pandora's box, and regulation may skew perspective, but won't be putting the lid back on. Strange shit happens when you consolidate all human knowledge into 4 frontier models controlled by a small group of people, all of which are now deeply involved with the Pentagon. And the only substantial competition is Chinese... I don't think "interesting times" was ever intended to suggest this degree of intensity. Yep, I'm an alarmist when not leaving funny comments. I think I'll go listen to some Yampolskiy now.