“The U.S. proposed new export restrictions on advanced AI technology, creating a three-tier system for access to AI chips and models. 🎯 Tier 1 nations, including key allies like Japan, retain broad access. 🎯 Tier 2 nations, like Singapore, face strict caps without a security
https://x.com/DeepLearningAI/status/1880341893873033511
“Fantastic to see the UK articulate a coherent, broad strategy for ensuring it leads on AI adoption and development. Lots of good ideas – AI growth zones, unlocking national data, 20X’ing public compute. Also v excited to see emphasis on funding technical regulators!” / X
https://x.com/jackclarkSF/status/1878821057370681466
“🚀 UK to ‘unleash’ AI to turbocharge economy Has unveiled an ambitious AI Opportunities Action Plan aimed at boosting economic growth supported by £14 billion in investments and 13,250 new jobs from tech firms. → The action plan emphasizes productivity improvements in public
https://x.com/rohanpaul_ai/status/1878802137674916275
“DeepSeek-V3, the company’s latest open LLM, surpasses Llama 3.1 405B and GPT-4o on key benchmarks, especially in coding and math tasks. Using a mixture-of-experts architecture with 671 billion parameters, only 37 billion are active at once, DeepSeek V3 was trained at a low cost” / X
https://x.com/DeepLearningAI/status/1880087643964199260
“Or impact of the fact that closed-source players use open-source AI (currently dominated by Chinese players) like open-source datasets? The countries or companies that win open-source AI will have massive power and influence on the future of AI.” / X
https://x.com/ClementDelangue/status/1877767382120255792
“needs a vibe check but the paper looks very solid to me. Another current-gen Chinese MoE design, and the first open model with frontier-class long context support. 1/8 lightning [or other accelerated mixer]/softmax attention mix is about what I’d expect from V4. It has began.
https://x.com/teortaxesTex/status/1879251029084225966
“Process Reward Models (PRM) and online RLHF are the unhidden secret to creating reasoning models like o1. PRMs are used to identify and mitigate intermediate reasoning errors in the Chain-of-Thought. The @Alibaba_Qwen Team just released “Lessons of Developing Process Reward
https://x.com/_philschmid/status/1879101437663154194
“NeuroXess reported two advances in BCI, allowing brain-injured patients to operate robots and communicate through thought In addition, the BCI also achieved to the first real-time decoding of Chinese language from brain signals
https://x.com/adcock_brett/status/1878488077775028676
OpenAI’s AI reasoning model ‘thinks’ in Chinese sometimes and no one really knows why | TechCrunch
OpenAI’s AI reasoning model ‘thinks’ in Chinese sometimes and no one really knows why
(2) Deepseek: The Quiet Giant Leading China’s AI Race
https://www.chinatalk.media/p/deepseek-ceo-interview-with-chinas
“Restricting mainstream AI tech won’t strengthen security but risks eroding U.S. global competitiveness and innovation. 🚨 NVIDIA’s statement criticizes the Biden Administration’s “AI Diffusion” rule. → NVIDIA highlights the legacy of the Trump Administration’s approach, which
https://x.com/rohanpaul_ai/status/1878804099053027430
Nvidia stock falls after Biden administration releases updated export rule for AI chips
https://finance.yahoo.com/news/nvidia-stock-falls-after-biden-administration-releases-updated-export-rule-for-ai-chips-151755666.html?guccounter=1
ByteDance appears to be skirting US restrictions to buy Nvidia chips: Report | TechCrunch
ByteDance appears to be skirting US restrictions to buy Nvidia chips: Report
OpenAI urges US to prioritize AI funding, regulation to stay ahead of China | Reuters
https://www.reuters.com/technology/artificial-intelligence/openai-urges-us-prioritize-ai-funding-regulation-stay-ahead-china-2025-01-13/
