[AINews] 10% worse, 100x cheaper, 10000x faster: Why Simulation is taking over

SummaryAI generated

According to Latent Space, simulation technology is rapidly replacing traditional methods. This technology reduces performance by about 10% compared to existing solutions, but costs are reduced by 100 times, and operating speed is increased by 10,000 times.

Related eventsRELATED EVENTS

2026-09-07 21:02The `GPT-6 Astra model was released, and OpenAI President Brockman confidently declared, “Welcome to the era of AGI.”
2026-09-08 19:21OpenAI released GPT-6 Astra, marking the arrival of the AGI era, with capabilities for autonomous computer operation and scientific research.
2026-09-08 22:40① Nvidia CEO Jensen Huang posted that OpenAI’s GPT-6 Astra, released last week, was trained using approximately 100,000 NV Link 72 clusters, and he believes that General Artificial Intelligence (AGI) has officially arrived; ② GPT-6 Astra can directly operate computers and software to perform complex tasks such as programming, reaching the most advanced level in multiple fields. OpenAI has announced the beginning of the AGI era.

Key entitiesKEY ENTITIES

Constitutional AIInstructGPTLee et al.MicrosoftPhi seriesPoolsideZ.ai GLMphi-1.5

Coverage · reports per dayLANGUAGE SPLIT

Entity relations

SignalsSIGNALS

Keyword heat

Z.ai GLM1
Poolside1
InstructGPT1
Constitutional AI1
Lee et al.1
Microsoft1
Phi series1
phi-1.51

All reports (1)SOURCES

↗

L Latent Space en 2026-08-22 15:36

Since 2022, machine intelligence has added one component per year that is manufactured by models rather than humans. This trend is characterized by “human simulation,” where the generated content performs approximately 10% worse than human performance, but at a cost 100 times lower and with a speed increased by 10,000 times. This process consists of five stages: in 2022, LLMs replaced human judgment with rewards; in 2023, training data became synthetic materials and web page rewrites generated by LLMs; in the same year, teacher models cloned the behavior of cutting-edge models through instruction fine-tuning; in 2024, model-driven task selection and evaluation were implemented for course generation; and by 2026, the role of researchers was gradually replaced by AI assistants such as Copilot and SWE-agent.