Everyone tells you to practice speaking from day one.
That's backwards. Read first, talk later.
Your brain needs input before you can say much. Get thousands of words in by reading.
This is called comprehensible input, Stephen Krashen's method and ALG (Automatic Language Growth) both show it works.
Start with graded readers at your level. Then move to simple news articles. Finally tackle native content.
Spend 15 minutes daily reading to make good progress.
Read for 2-3 months, then start speaking. You'll sound way better.
Guido retweeted
I don't understand why people don't just lock in on YouTube
My faceless YouTube channels make me $120,000+/month (long form only)
There is 𝗻𝗼 𝗹𝘂𝗰𝗸 involved.
Let me send you a free course on exactly how to do it.
To get: -
1. Follow (So I can DM you )
2. Retweet
3. Reply " Yt " Must follow me to get DM.
Free for 48 hours.
Guido retweeted
Last chance to turn it off.
On Monday, November 3rd, Microsoft will start using your LinkedIn data for AI training. And remember, you're opted in by default.
To toggle it off 👉 Account - Settings & Privacy > Data privacy > Data for Generative AI Improvement.
Thailand’s 2025 tourism crash explained – empty beaches, rising costs, Chinese tourists gone. Here’s what’s happening 🧵
1
1
95
6️⃣ The real blow: China pulled out
– Hit movie No More Bets sparked fear of Southeast Asia
– Celebrity kidnapping in Thailand confirmed worst fears
🚫 Chinese tourists = GONE
1
1
95
The newest @lexfridman podcast about AI has great insights. To save 5 hours of your life, these are the key insights:
1. NVIDIA’s CUDA Lock-In Is a Major Barrier for Competitors
• CUDA’s maturity and developer support give NVIDIA a huge advantage, making it difficult for competitors like AMD or Huawei to catch up. But companies like DeepSeek and China are trying to bypass this through alternative low-level programming methods.
2. China Is Rapidly Developing Domestic AI Chips
• With restrictions on GPU exports, China has accelerated the development of its own high-performance chips (Biren BR100, Huawei A910). These chips are already being deployed in AI cloud hosting and inference workloads.
3. DeepSeek’s Cost Efficiency Comes from Hardware-Software Co-Design
• DeepSeek’s success lies in programming optimizations at a low level, which maximizes GPU utilization and reduces training costs. They’ve engineered custom kernels to bypass some inefficiencies seen in traditional models like OpenAI’s.
4. Export Controls on GPUs Could Change the AI Race
• The U.S.’s restrictions on exporting advanced GPUs (like NVIDIA A100 and H100) to China are intended to slow down China’s AI development, but domestic solutions are emerging faster than expected.
5. AI Training Costs Are Dropping Fast
• Thanks to optimizations like model pruning, quantization, and distributed training, the cost of training large models is falling. This makes smaller AI labs and startups more competitive against giants like OpenAI and DeepMind.
6. Taiwan’s TSMC Is a Strategic Asset in the AI Race
• TSMC, the world’s leading semiconductor manufacturer, is critical to producing advanced AI chips. If geopolitical tensions around Taiwan escalate, it could disrupt the global supply of high-end chips, affecting companies like NVIDIA and AMD.
7. AI Clusters Are Scaling Rapidly (AI Megaclusters)
• Companies like DeepSeek, OpenAI, and DeepMind are building massive AI megaclusters to train next-gen models. These clusters use thousands of GPUs or custom chips working in parallel to push the limits of model size and complexity.
8. AGI Timelines Are Uncertain, But Progress Is Rapid
• The video likely discusses how predicting AGI timelines is tricky, but advances in large language models (GPTs) and reinforcement learning breakthroughs (like AlphaGo) suggest we are getting closer.
9. AMD Could Close the Gap If DeepSeek Optimizes for Non-NVIDIA Chips
• DeepSeek’s use of lower-level programming layers (like PTX) could make their models easier to port to AMD’s GPUs. If this happens, NVIDIA’s dominance in AI hardware could face serious competition.
10. AI Is Central to U.S.-China Geopolitical Rivalry
• AI isn’t just about innovation—it’s also a national security priority. The U.S. views control of AI technology as crucial, which is why export controls, espionage, and chip production are highly politicized.
The podcast:
teapawt.net/lexfridman/status/1886…
Here's my 5-hour conversation with @dylan522p and @natolambert on DeepSeek, China, OpenAI, NVIDIA, xAI, Google, Anthropic, Meta, Microsoft, TSMC, Stargate, megacluster buildouts, RL, reasoning, and a lot of other topics at the cutting edge of AI. This is was a mind-blowing, super-technical, and fun conversation.
Yes, we discuss r1 and o3-mini, but more importantly we look into the future of technology, geopolitics, and humanity in a world that stands on the precipice of a global AI revolution.
The first 4 hours are here on X (4 hours is current limit), and the full 5 hours are up everywhere else. Links in comment.
Timestamps:
0:00 - Introduction
3:33 - DeepSeek-R1 and DeepSeek-V3
25:07 - Low cost of training
51:25 - DeepSeek compute cluster
58:57 - Export controls on GPUs to China
1:09:16 - AGI timeline
1:18:41 - China's manufacturing capacity
1:26:36 - Cold war with China
1:31:05 - TSMC and Taiwan
1:54:44 - Best GPUs for AI
2:09:36 - Why DeepSeek is so cheap
2:22:55 - Espionage
2:31:57 - Censorship
2:44:52 - Andrej Karpathy and magic of RL
2:55:23 - OpenAI o3-mini vs DeepSeek r1
3:14:31 - NVIDIA
3:18:58 - GPU smuggling
3:25:36 - DeepSeek training on OpenAI data
3:36:04 - AI megaclusters
4:11:26 - Who wins the race to AGI?
4:21:39 - AI agents
4:30:21 - Programming and AI
4:37:49 - Open source
4:47:01 - Stargate
4:54:30 - Future of AI
3
110
Curious about how modern scammers manipulate trusted tech services to steal from unsuspecting victims? Exclusive Insights from Brian Krebs: A Deep Dive into a Prolific Voice Phishing Operation
1
82
One of the most effective tactics used by these criminals is sending fake Apple or Google security alerts, followed by a call from a spoofed number claiming to be tech support.
1
75
His latest article is a must-read for anyone interested in cybersecurity and the evolving methods of cybercriminals.
krebsonsecurity.com/2025/01/…
61









