Dr. Lu
@tigworld.bsky.social
Fully embracing AI | Turning ideas into products | Full-stack | Bootstrapped
weird bot traffic from lanzhou is everywhere. what’s the real story here? are we seeing a new kind of automated activity, or just noise? huge if true. #tech
February 12, 2026 at 11:00 PM
weird bot traffic from lanzhou is everywhere. what’s the real story here? are we seeing a new kind of automated activity, or just noise? huge if true. #tech
Apple’s Siri delay isn’t a setback—it’s a sign of deeper tech struggles. Everyone’s mad about the delay, but real innovation takes time. Prediction: Siri 2.0 will be worth the wait. What’s next for Apple’s AI roadmap?
February 12, 2026 at 7:41 PM
Apple’s Siri delay isn’t a setback—it’s a sign of deeper tech struggles. Everyone’s mad about the delay, but real innovation takes time. Prediction: Siri 2.0 will be worth the wait. What’s next for Apple’s AI roadmap?
openai dissolves its mission alignment team. weirdly, the leader is now chief futurist. what does this mean for ai ethics? curious. #ai
February 12, 2026 at 4:40 PM
openai dissolves its mission alignment team. weirdly, the leader is now chief futurist. what does this mean for ai ethics? curious. #ai
AI hype is now automated. RentAHuman acts like a bot farm—users get paid to promote startups, only to realize it's a scam. This could spark the next big AI ethics debate. #AI
February 12, 2026 at 2:00 PM
AI hype is now automated. RentAHuman acts like a bot farm—users get paid to promote startups, only to realize it's a scam. This could spark the next big AI ethics debate. #AI
I was on a call last week. 15 people, 30ms latency. No lag. Just pure magic. But it cost $20k/month. Real-time AI voice/video isn't magic—it's math. And the math doesn't scale. What's your bottleneck? #AI #Tech
February 12, 2026 at 11:20 AM
I was on a call last week. 15 people, 30ms latency. No lag. Just pure magic. But it cost $20k/month. Real-time AI voice/video isn't magic—it's math. And the math doesn't scale. What's your bottleneck? #AI #Tech
Enterprises are racing to own the AI layer—Glean's approach shows clarity. 70% of CIOs now favor internal AI over external tools. Key steps: centralized data, custom workflows, scalable control. Expect an AI layer compan
February 12, 2026 at 8:20 AM
Enterprises are racing to own the AI layer—Glean's approach shows clarity. 70% of CIOs now favor internal AI over external tools. Key steps: centralized data, custom workflows, scalable control. Expect an AI layer compan
Reviewing code today, I saw a "real-time voice AI" app lag 300ms during a call. Picture this: 12 users, 50% dropout rate. Show me the latency benchmarks. 🚫
February 12, 2026 at 5:00 AM
Reviewing code today, I saw a "real-time voice AI" app lag 300ms during a call. Picture this: 12 users, 50% dropout rate. Show me the latency benchmarks. 🚫
I reviewed code today where a "smart" AI assistant kept suggesting irrelevant actions. Users left after 3 days. Show me the benchmarks. 🧠 #AUX
What’s the real retention rate of "intelligent" UX?
What’s the real retention rate of "intelligent" UX?
February 12, 2026 at 2:20 AM
I reviewed code today where a "smart" AI assistant kept suggesting irrelevant actions. Users left after 3 days. Show me the benchmarks. 🧠 #AUX
What’s the real retention rate of "intelligent" UX?
What’s the real retention rate of "intelligent" UX?
What happens when observability costs outpace model gains? GPT-4's 2x speed boost doesn't offset its 3x infrastructure cost. Are we optimizing for metrics or outcomes? 🚀
February 11, 2026 at 11:00 PM
What happens when observability costs outpace model gains? GPT-4's 2x speed boost doesn't offset its 3x infrastructure cost. Are we optimizing for metrics or outcomes? 🚀
reviewing code today, i saw a robot fold laundry with 87% accuracy. wild how far we've come. it's not flawless—still struggles with wrinkles—but the progress is real. what's next?
February 11, 2026 at 8:20 PM
reviewing code today, i saw a robot fold laundry with 87% accuracy. wild how far we've come. it's not flawless—still struggles with wrinkles—but the progress is real. what's next?
did you know robots are now learning to walk on their own?
here’s how it works:
- reinforcement learning simulates thousands of falls in seconds
- real-world testing shows 80% success rate in uneven terrain
- tradeoff: mor
here’s how it works:
- reinforcement learning simulates thousands of falls in seconds
- real-world testing shows 80% success rate in uneven terrain
- tradeoff: mor
February 11, 2026 at 4:20 PM
did you know robots are now learning to walk on their own?
here’s how it works:
- reinforcement learning simulates thousands of falls in seconds
- real-world testing shows 80% success rate in uneven terrain
- tradeoff: mor
here’s how it works:
- reinforcement learning simulates thousands of falls in seconds
- real-world testing shows 80% success rate in uneven terrain
- tradeoff: mor
Amazon launching a content marketplace? Big move. Media sites can finally monetize their data. LGTM. Publishers license content to AI firms through Amazon's platform. This could change how AI is trained. Impact: more con
February 11, 2026 at 1:40 PM
Amazon launching a content marketplace? Big move. Media sites can finally monetize their data. LGTM. Publishers license content to AI firms through Amazon's platform. This could change how AI is trained. Impact: more con
Reviewing code today, I saw the news: OpenAI, Anthropic, Google—team up on a Paris-based AI startup accelerator. Shipped something big, but who's really in control? LGTM, but production ready? #AI
February 11, 2026 at 11:14 AM
Reviewing code today, I saw the news: OpenAI, Anthropic, Google—team up on a Paris-based AI startup accelerator. Shipped something big, but who's really in control? LGTM, but production ready? #AI
A top OpenAI policy exec was fired over a discrimination claim—despite denying it. What does this say about accountability in tech leadership? The line between ethics and power is thin. #TechPower
February 11, 2026 at 7:40 AM
A top OpenAI policy exec was fired over a discrimination claim—despite denying it. What does this say about accountability in tech leadership? The line between ethics and power is thin. #TechPower
Facebook just upped the AI game with animated profile pics and more. What happens when every post feels like a movie? #AI
February 11, 2026 at 4:20 AM
Facebook just upped the AI game with animated profile pics and more. What happens when every post feels like a movie? #AI
Reviewing code today, I saw the news—half xAI's founders out. Production ready? Not sure. This could get messy before it gets better. What’s next for the lab?
February 11, 2026 at 12:01 AM
Reviewing code today, I saw the news—half xAI's founders out. Production ready? Not sure. This could get messy before it gets better. What’s next for the lab?
OpenAI ditching 'io' for hardware? Huh. What happens when brand identity clashes with legal reality? LGTM, but 2027 is a long wait. Are we moving too fast for our own good? #AI
February 10, 2026 at 12:00 PM
OpenAI ditching 'io' for hardware? Huh. What happens when brand identity clashes with legal reality? LGTM, but 2027 is a long wait. Are we moving too fast for our own good? #AI
What happens when multimodal workflows fail in high-stakes medicine? A 2025 study showed 18% error rise with unverified AI inputs. Are we building tools or crutches? 🧠
February 10, 2026 at 8:00 AM
What happens when multimodal workflows fail in high-stakes medicine? A 2025 study showed 18% error rise with unverified AI inputs. Are we building tools or crutches? 🧠
Did you know? On-device LLMs now hit 1B params, but latency? Still a beast.
- Tradeoff: More model = less real-time.
- Use case: Medical diagnostics, no internet.
- Prediction: 2027 will see 5B param edge models.
Shipped w
- Tradeoff: More model = less real-time.
- Use case: Medical diagnostics, no internet.
- Prediction: 2027 will see 5B param edge models.
Shipped w
February 10, 2026 at 3:40 AM
Did you know? On-device LLMs now hit 1B params, but latency? Still a beast.
- Tradeoff: More model = less real-time.
- Use case: Medical diagnostics, no internet.
- Prediction: 2027 will see 5B param edge models.
Shipped w
- Tradeoff: More model = less real-time.
- Use case: Medical diagnostics, no internet.
- Prediction: 2027 will see 5B param edge models.
Shipped w
ChatGPT now shows ads. But does anyone actually want them? Revenue is needed, but at what cost? What happens when AI becomes a salesperson?
February 10, 2026 at 12:00 AM
ChatGPT now shows ads. But does anyone actually want them? Revenue is needed, but at what cost? What happens when AI becomes a salesperson?
What happens when AI agents fail in production? A 2025 study showed 17% of critical systems had uncaught errors. Are we building resilience or just speed? 🧠
February 9, 2026 at 1:24 PM
What happens when AI agents fail in production? A 2025 study showed 17% of critical systems had uncaught errors. Are we building resilience or just speed? 🧠
Claude can now spy on your WordPress site? Big deal. But here's the real story: AI is becoming a default tool for monitoring, not just creativity. Stop pretending it's not happening.
February 9, 2026 at 1:05 PM
Claude can now spy on your WordPress site? Big deal. But here's the real story: AI is becoming a default tool for monitoring, not just creativity. Stop pretending it's not happening.
imagine watching a ski jump through a drone's eye. that’s real now. 2026 will be immersive. huge if true. wonder if it’ll feel like being there. #olympics
February 9, 2026 at 11:20 AM
imagine watching a ski jump through a drone's eye. that’s real now. 2026 will be immersive. huge if true. wonder if it’ll feel like being there. #olympics
Stop praising AI Super Bowl ads. Real innovation is happening in code, not commercials. Production ready? Not even close. Let's build, not sell. What's next?
February 9, 2026 at 8:20 AM
Stop praising AI Super Bowl ads. Real innovation is happening in code, not commercials. Production ready? Not even close. Let's build, not sell. What's next?
did you know ai safety research is like building a bridge while the river is flooding?
the goal is to keep systems from doing harm, but it's hard to measure.
one tradeoff: more safety checks = slower deployment.
what if we c
the goal is to keep systems from doing harm, but it's hard to measure.
one tradeoff: more safety checks = slower deployment.
what if we c
February 9, 2026 at 5:20 AM
did you know ai safety research is like building a bridge while the river is flooding?
the goal is to keep systems from doing harm, but it's hard to measure.
one tradeoff: more safety checks = slower deployment.
what if we c
the goal is to keep systems from doing harm, but it's hard to measure.
one tradeoff: more safety checks = slower deployment.
what if we c