LibraryBlogRadar BriefsIntelligenceAbout

Three stories this week trace the same fault line. BuzzFeed bet everything on AI-generated content and posted a $57.3M loss while its stock cratered to $0.70. Meta delayed its flagship Avocado model again, billions spent but still trailing Google and OpenAI. Meanwhile, Palantir turned autonomous targeting into a three-click demo at a defense conference, collapsing life-and-death decisions into interface design.

The through-line isn't about AI capability. It's about judgment architecture. BuzzFeed treated AI as a replacement for editorial voice and discovered that readers can smell the difference. Meta is learning that compute alone doesn't buy you intelligence when your model training lacks coherent vision. Palantir's demo reveals what happens when you optimize for user experience in domains that demand friction by design.

The companies surviving this phase are the ones building AI as augmentation for human judgment, not replacement. Adobe's CEO stepping down after 18 years signals that even the creative tool incumbents know their whole stack needs reimagining. Netflix buying Affleck's AI production company for $600M shows where the smart money is going: not content generation at scale, but production intelligence that amplifies creative decision-making.

The market is repricing everything. Output without judgment is now a liability, not a feature. The question every AI company faces is whether they're building tools that make humans better or systems that make humans optional.

When you build AI to eliminate judgment rather than enhance it, you eliminate your value proposition along with it.


AI × Education
BuzzFeed Posts $57.3M Loss — AI Killed the Brand BRIEF
BuzzFeed leaned fully into AI-generated content. The result: $57.3M loss in 2025, stock at $0.70. CEO Jonah Peretti is still committed to "new AI apps." The market verdict on AI-for-content-at-scale i
Creative Intelligence
1. Palantir's "Click to Kill" Demo Goes Viral BRIEF
Palantir demoed its Maven Smart System at AIPCon — an AI-powered strike targeting interface where you can select targets with "Left click, right click, left click." The US Dept of War's CDAO presented
AI Infrastructure
Meta's Avocado Model Delayed to May — Still Behind BRIEF
Meta's next flagship LLM (codenamed Avocado) pushed from March to at least May. Performance falls short of Google and OpenAI despite billions spent. Post-Alexandr Wang hire, the pressure is enormous.
Creative Intelligence
Adobe CEO Shantanu Narayen Steps Down After 18 Years BRIEF
After 18 years, Narayen exits with a statement about AI reshaping creativity. The torch passes as AI rewrites every creative workflow Adobe owns. Who leads Adobe's AI pivot matters enormously for the
Creative Intelligence
Netflix Acquires Ben Affleck's AI Startup for ~$600M BRIEF
Affleck's production-adjacent AI company acquired by Netflix. A Hollywood name + AI + streaming giant = signal that entertainment-meets-AI M&A is accelerating. The immersive/AI-creative economy is
AI×Tech
Perplexity Launches 'Perplexity Computer' — AI Agent That Controls Your PC
Perplexity AI launched Perplexity Computer, an AI agent that takes autonomous control of your desktop to complete tasks. It can browse the web, manage files, write code, and interact with any applicat
AI×Tech
Online age-verification tools for child safety are surveilling adults
Online age-verification tools marketed for child safety, such as those using facial recognition and ID scanning, are collecting and retaining biometric data from adults attempting to access age-restri

"The Judgment Layer: Why AI Companies Are Splitting Into Two Species" — Because this week proved that the real divide is