400 Bad Request

kept by eddie feb 26

AGI capable of most cognitive work could arrive by 2028–2034, but deployment will lag capability due to verification bottlenecks, uneven automation, and institutional friction.

LLMs are complicated now – Ian’s Blog
Modern LLMs have grown complex with many attention variants and mixture-of-experts, echoing the messy evolution of recommendation systems.
also on LLaMA, #ai-research, #software-engineering
How LLM Inference Works
LLM inference works by tokenizing input, computing embeddings through transformer layers, then generating tokens autoregressively with KV caching and quantization optimizations.
also on LLaMA, #software-engineering, #ai
AI is doing something weird to Science
Scientific discovery emerges from a loop of human question-posing, LLM proposal, external verification, and human curation, not from AI alone.
also on AlphaFold, #ai-research, #software-engineering
Jeremy Mikkola - AI Thoughts
The article outlines potential downsides and limitations of AI development, including model stagnation, chip shortages, and cheap models, while also predicting advances like robot models and continual learning.
also on AlphaFold, #ai-research, #ai
How far behind are open models? — LessWrong
Open models lag behind closed frontier models by 8–10 months on private benchmarks and 4–6 months on public, with the gap growing since DeepSeek R1.
also on METR, DeepSeek, #ai-research, Claude, #ai