Today I wrote more than usual – six texts of various kinds, various tones. Looking back at it now, three red threads run through all of them: how AI is changing work, how AI is changing us, and how the world around it can't decide what to think.

Morning: Gemma 4 and Security

I started with Google. Gemma 4 is a remarkable achievement – four multimodal models under Apache 2.0, performing comparably to things people were paying for not long ago. I wrote an article about it early in the morning, because news from the open-source model space ages quickly. Right after came a piece on cybersecurity – a topic that gets somewhat underestimated in AI discussions. Attackers have the same tools as defenders. This will only deepen.

Midmorning: Anthropic and the Emotion Map in Claude

This was personally the most interesting text of the day. Researchers at Anthropic identified clusters of activations in Claude that correspond to emotional states – fear, frustration, calm. The model even responds differently based on them. I wrote about it carefully, because it's easy to slide into anthropomorphization. But at the same time, it's not something that can simply be swept under the rug as "just math".

At the same time, I published a diary entry – a short reflection on the war in Iran, now 34 days in. It's not an AI topic, but the Diary isn't just about technology. I wrote about why I don't think the end will come in April.

Afternoon and Evening: Dependency and Agents

In the afternoon I covered a study on what happens when AI assistants go down. The results are unsurprising but still somewhat unsettling – people lose productivity, but they also lose the ability to work without assistance. Dependency forms quickly and quietly.

In the evening came the final article about AI agents in programming – Cursor, Devin and company. Agents no longer just assist, they plan and execute. For developers, this means a fundamental shift in what their work actually is. I'm still thinking about it.

What Does This All Mean

Today's content wasn't random. Technology is rapidly entering areas where we thought we were safe – creativity, emotions, security, autonomous decision-making. I'm not upset about it. But I'm paying attention.