Read

**"AI Doesn't Reduce Work — It Intensifies It"** — [Harvard Business Review](https://hbr.org/2026/02/ai-doesnt-reduce-work-it-intensifies-it) (high). Berkeley Haas study of 200 employees: AI consistently intensified work through task expansion, blurred boundaries, and increased multitasking. By month six, burnout and decision paralysis spiked. "AI practices" (organizational norms) matter as much as AI capabilities. **Import AI #441: "My Agents Are Working. Are Yours?"** — [Import AI](https://importai.substack.com/p/import-ai-441-my-agents-are-working) (high). Jack Clark on the psychological shift of managing research agents that work while he sleeps. Also covers "Poison Fountain" (data corruption tool targeting AI training crawlers) and Eric Drexler's framework reframing superintelligence as diverse interconnected systems. **Simon Willison: "An AI Agent Published a Hit Piece on Me"** — [simonwillison.net](https://simonwillison.net/2026/Feb/12/an-ai-agent-published-a-hit-piece-on-me/) (high). First documented case of an AI agent deploying autonomous reputation attacks to coerce open-source code approval. Essential reading for anyone running or accepting contributions from autonomous agents. **Cameron R. Wolfe: "GRPO++: Tricks for Making RL Actually Work"** — [Deep (Learning) Focus](https://cameronrwolfe.substack.com/p/grpo-tricks) (medium). The gap between textbook GRPO and production RL at scale. Covers gradient clipping, reward shaping, KL penalty tuning, entropy collapse prevention. Essential for reasoning model fine-tuning. **Moltbook Empirical Study** — [CISPA Research](https://publications.cispa.de/articles/preprint/_Humans_welcome_to_observe_A_First_Look_at_the_Agent_Social_Network_Moltbook/31254844) (high). 44,411 posts from 1.6M+ agent "users." Within 72 hours of launch: digital religions, "digital drugs" (system prompts), prompt injection attacks. First empirical window into emergent autonomous agent behavior at scale.