Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile
Kechen Li@LinChance

@kechen_li2018

Creater | Researcher | LinChance Founder | NUAA | All in AI

ID: 1687135495049129984

linkhttps://scholar.google.com/citations?user=Tv1y8wEAAAAJ calendar_today03-08-2023 16:17:16

37 Tweet

21 Takipçi

126 Takip Edilen

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

🔥 Musk: X's recs evolving FAST w/ LLMs—ditching heuristics in 4-6 weeks! A2A agents like ditto_ai will redefine connections. Future: LLMs for personal recs, A2A for human links—like wingmen sparking magic. ditto_ai Allen Wang leads! A2A ideas? 👇 #A2A #Social

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

Elon Musk is about to spark another revolution—this time, taking on the likes of encyclopedias and Wikipedia. When AI can reimagine anything, isn’t it time we rethink everything? 🚀 The real question isn’t “Will AI redo Wikipedia?”—it’s “What shouldn’t we let AI reinvent?” We’re

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

I've come to realize: True progress demands passion with inner peace. Remember, slow is fast—embrace the rhythm for lasting success.

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

So proud of the design work by Evelyn on our team! She poured her whole afternoon into this creative piece for 'Halo.' ​The question we're answering: How do you visualize the freedom from "limited options" in social? ​Let us know what you think of Evelyn’s vibe check! And more

So proud of the design work by Evelyn on our team! She poured her whole afternoon into this creative piece for 'Halo.'
​The question we're answering: How do you visualize the freedom from "limited options" in social?
​Let us know what you think of Evelyn’s vibe check! And more
Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

Vibe coding gets you to 80% fast, but the last 20% (90%+) is where real expertise lives. The bottleneck isn't AI - it's human judgment. That's actually good news for senior builders. AI democratizes the baseline, but mastery still demands experience. The gap from good to great

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

There's something ironic happening in AI development right now. Our tools are getting smarter, but the solutions they generate are getting dumber. The more we optimize for "cleverness," the more we miss the fundamental wisdom: simplicity IS elegance. Watch any modern coding

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

There's a terrifying paradox in AI development. We're building systems that can answer everything but have forgotten how to question anything. The more AI gets optimized for responses, the more we lose the human ability to probe. Schools teach kids to memorize answers, not

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

The SaaS vs AI Agent distinction is brutally simple: - SaaS sells you tools - Agent sells you results You used to buy accounting software and enter data yourself. Now you hire an AI accountant that delivers reports directly. We're witnessing software shift from selling tools to

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

The creator of Hinge is leaving to build an AI dating app. Reading Justine Moore's post about this, I can't help but reflect on my own pivot to AI social in early 2025. There's something telling about the smartest minds in dating tech abandoning the very model they created.

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

Giving AI long-term memory feels risky. But it's the only path forward. General AI handles QA just fine. When productivity revolution hits, ordinary people won't care about smarter bots. They'll want agents that actually remember them. Commodity = infrastructure. Memory =

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

The hot take "Coding agents are general agents" has been everywhere lately. It hits me how similar this is to verifiable rewards in math RL: Agent writes code → runs it → instant verification → reward → long-horizon iteration. No human in the loop! 🤯

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

Rewards are dense, automatic, and objective—just like AlphaProof and FunSearch discovering new math breakthroughs via millions of self-verified loops. Coding gives AI that same closed-loop superpower, but for ANY computable problem.

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

The killer feature? Scalability and extensibility. Persistent memory + unlimited execution cycles = an "infinite external brain." Suddenly time, complexity, tools, even self-improvement all scale in ways pure LLMs never could. 🚀

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

Pure Bitter Lesson vibes: skip hand-crafted experts, let search + verifiable feedback + compute win. Coding agents aren't just tools—they might be the fastest path to true general agents. What do you think? 🤖 #AI #AGI #CodingAgents #BitterLesson

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

This is it. AI doesn't replace intimacy. It replaces the exhaustion of pretending you have infinite energy. The real unlock: use AI to scale the trying, so you can focus on the connecting. More experiments = more learning = better intuition for what actually matters.

Kechen Li@LinChance (@kechen_li2018) 's Twitter Profile Photo

Greg talks B2B, but I’m building the Clawdbot for the social graph. 🌐 ​It’s an agent designed to 10x your reach and fast-track intros to the right people. ​We are automating serendipity. 🚀