three AI reads at a time

rlft

2 items

WIRED 2026-05-10-2

I Work in Hollywood. Everyone Who Used to Make TV Is Now Secretly Training AI

Mercor's 300 employees plus tens of thousands of contractors is structurally identical to Medvi's 2 employees plus outsourced clinical labor — same shape, different industry. The frontier labs' "human alignment" premium is a labor-supply-chain bet, and procurement DD that asks about training-data provenance but not evaluation-labor provenance is asking 2024's question. The atomization Fowler describes is the durable feature: profession unbundled into rate-this, classify-that, evaluate-that, with the person erased and the signal extracted.

# tags
ai-labor-displacementai-economicsai-1.0-defensibilityai-vendor-governanceai-regulatory-riskai-training-as-infrastructureevaluation-infrastructureagent-supply-chainworkforce-bifurcationai-procurementtraining-datarlftwiredai-political-economyfilm-industryregulated-ai
◆ entities
MercorOutlierTuringScale AISurgeWIREDRuth FowlerWGASAG-AFTRAAppenMedviEU AI ActCalifornia ABC testPlatform Work DirectivePartnership on AIFairwork
→ threads
ai-labor-displacementai-1.0-defensibilityevaluation-infrastructureai-vendor-governance
⟷ links
2026-04-05-12026-03-18-12026-04-01-32026-04-16-12026-03-25-12026-04-07-12026-04-14-22026-04-26-32026-04-27-32026-05-01-32026-05-02-12026-05-04-32026-05-06-32026-05-08-32026-05-08-2
permalink
a16z Podcast (originally Cheeky Pint) 2026-04-17-3

From Models to Mobility: Waymo Architecture at Scale — Dolgov on the Teacher/Simulator/Critic Triad and the End-to-End Debate Resolution

Waymo's architecture resolves the end-to-end debate: Dolgov states pure pixels-to-trajectories drives "pretty darn well" in the nominal case but is "orders of magnitude away" from what full autonomy requires. The 500K-rides-per-week stack is one off-board foundation model fanning into three specialized teachers (Driver, Simulator, Critic), each distilled into smaller in-car students; RLFT against the critic is the physical-AI analog to RLHF. Enterprise teams shipping pure-LLM agents without the simulator and critic scaffolding are replaying Waymo's 2017, not its 2026: evaluation infrastructure is the reliability gate, not model choice.

# tags
physical-aiagentic-ai-viabilityai-1.0-defensibilityvertical-integrationreliabilityevaluation-infrastructurepilot-to-scalefoundation-modelswaymoautonomous-vehiclesrobotaxiai-architecturesensor-fusionend-to-end-learningrlftteacher-student-distillationalphabettesla-fsda16zpodcast
◆ entities
WaymoDmitri DolgovAlphabetTesla FSDDeepMindJaguar I-PACEHyundai IonicJohn CollisonGemini Robotics-ERPhysical Intelligence
→ threads
agentic-ai-viabilityai-1.0-defensibilityreliabilitypilot-to-scalevertical-integration
⟷ links
2026-04-03-22026-04-12-22026-04-10-12026-04-15-12026-03-23-12026-03-26-22026-03-30-32026-04-07-22026-03-27-22026-04-06-12026-04-11-12026-04-12-32026-04-13-32026-04-13-12026-04-15-2
permalink