AI Trend Radar: What's Hype, What's Deployed

Narratives (themes) are extracted from our continuously monitored AI-news stream by an LLM classifier, normalized, and admitted to the radar once they appear in at least 3 items in the 45-day window. Each narrative is also classified against a stable topic-kind schema (model, technique, infrastructure, market/capital, policy…) by majority vote across its items — the schema's categories stay fixed over time so kind-level trends remain comparable across snapshots, and quadrants derive from them. Each news item additionally carries an event type (launch, funding, acquisition, adoption…), shown as the blip's event mix. Individual companies, products, and use-case types appear only when they dominate the conversation (at least 12 items with 6+ in the last 14 days) — shown as diamonds. A vendor's own newsroom never counts toward that vendor's dominance: only independent coverage does, and any own-channel share is disclosed on the blip.

The evidence ring comes from the use-case catalog: each news item is linked to published deployment cases by embedding similarity, but a case only counts toward the ring when at least two independent articles point to it (a single match is weak signal). A narrative's corroborated-case count places it — Deployed at scale (5+), In production (2+), Piloting (1+), or Discussion only (0). Most narratives sit in the outer discussion ring: that is the point — far more of the AI conversation is talk than is corroborated by real deployments. Marker size and label size reflect news volume and dominance (score = items + 2 × last-14-day items); color shows the dominant source class behind the narrative; ▲/▼ mark rising/cooling — shown only when there is enough volume and a real prior baseline to measure a direction, so most narratives read as steady rather than a fabricated trend. A dashed halo marks narratives new versus the previous daily snapshot; snapshots are kept per day so each narrative's history (sparkline + ring movements) builds automatically. Select any blip to see the headlines and cases behind it.