doug chang

nobody

[Kun Chen @kunchenguid]

overheard from friends talking about AI in China, and it’s very alarming

- AI short films went massive in China market because of how good the video models are now

- the first wave of people who mastered using AI to generate short films and comics made quite some money because they could produce a lot more content than traditional workflows

- but within just a few months, everybody and their mom could make equally good short films because all the expertise could be written in a markdown file

- when everyone’s more productive, no one actually has an advantage and it’s a race to the bottom. very quickly it became impossible to make money, and everyone’s busier than they were without AI

- the only one making money now is companies like bytedance who owns user demand and distribution. they sell ads to creators who pay to get traffic, and eventually ad spend became 80% of those creators’ operating expense, while AI tokens is less than 5%

so what are the problems here?

1. there’s no moat if you are creating anything with AI

if a human is really good at something, and it’s hard to find more humans who can do the same, they can leverage the scarcity of their skill to get paid very well

but if AI is good at something, it will be infinitely replicated, so there’s no scarcity hence you can’t really expect a sustainable business by creating anything with AI

2. more creators created more content people actually liked. that’s a good thing

but the economical benefit of that all got absorbed by the distribution platforms

there’s a famous chinese poem 遍身罗绮者，不是养蚕人 - it basically says people who dress well aren’t the people making dresses. that’s exactly what’s happening here

if our society doesn’t figure out a better way to distribute wealth, AI will help billionaires become trillionaires while average people become more exhausted than ever

[doug chang @dougc333]

@michaelryan207 @Diyi_Yang @jyangballin @stanfordnlp Hello Michael!! Your lectures are excellent. Riveting. I went to them personally. Hope to see you out there in the wild

[Matthew Pfab @MatthewPfab]

That’s not rain! Our coastal radar (KLGX) is picking up the massive waves along the WA coast! Due to its low beam tilt angle (0.0°), the radar is able to see the large waves (20-25 feet) and some of the spray they kick up! A unique quirk of PNW weather watching! #wawx

[doug chang @dougc333]

@JinWeiyang18434 I hope you recover quickly.

[doug chang @dougc333]

@rikiyatakehi @a1zhang @lateinteraction @jayden_teoh_ great idea

[Di Zhang @di_zhang_fdu]

What Is RLCD? The Secret Behind Jev
https://t.co/p5zKzGEqmP

[Sasha Dog (Hiring) 🫶🏼 @hackgoofer]

Weee. Finally it’s out!

Safety, certainty and trust > FUD

Been building TypeSafe for the last two years - excited to share with the community for early access. Please sign up at TypeSafe’s website.

[doug chang @dougc333]

@VolumeDynamics deepseek released a less memory architecture. https://t.co/RAxECLENLY

[tobi lutke @tobi]

Training tiny models for special purpose use cases works so incredibly well if you have a great self improving recursive flywheel. Shopify ML team is on fire.

finetuned 0.8b model beats GPT 5.6-sol xhigh in this very specialized task.

[Haoyang Weng @ElijahGalahad]

HDMI rewritten, now open-sourced at https://t.co/eU6qtJuHEV.

The video can be replayed by the same command README.md.

New features
1. mjlab backend
2. single object, multi motion support

... to be continued.

[steve hsu @hsu_steve]

Connectome is All You Need?

[Max Hodak @maxhodak_]

the single thing that would most accelerate our work at @ScienceCorp_ is a connectome

the mouse is probably only ~$300M away, something highly worth funding

[Edward Hu @edwardjhu]

How does one RL post-train a 397B model for long-horizon knowledge work? 👩‍💼

We share every step we took to bring Qwen 3.5 397B from 16.1% Pass@1 to 27.3% on APEX-Agents using DPPO, including final models weights and the full training script🚀 This is the first of many works from Mercor Research on open model training research.

Full blog: https://t.co/GWgR2DB7iy
Source code: https://t.co/96h8ADfvBp

[doug chang @dougc333]

@_xjdr https://t.co/n7RYdYHQIa can only get af far as tangent space :)

[doug chang @dougc333]

@_xjdr wow in d dimensions the sphere is 3d or N dims? Stunning work.

[Evergreen Intel @vcdgf555]

There is a major disaster unfolding along the Nepal-China border today.

Likely many deaths in this one video alone. Hundreds are missing.

[Alex Shaw @alexgshaw]

We're growing the @harborframework team!

If you're interested in agent evaluation and RL envs, please reach out.

A conservative estimate is that Harbor will process >1B rollouts and >$10B of token spend this year.

🧵 for more details

[Yuchen Jin @Yuchenj_UW]

UC Berkeley open-sourced FreeToken. Wild results:

A single RTX PRO 6000 runs the 753B GLM-5.2 at 14.9 tok/s!

An 8GB RTX 4060 laptop (~$1,000) runs Qwen3.6-35B at 39.3 tok/s!

FreeToken is 2–4x faster than Ollama across consumer GPUs. Local AI inference is getting very real. Great work by @Andy_ShuoYang and UC Berkeley Sky Lab!

[Jacky Kwok @jackyk02]

Scaling self-verification with DeepSeek V4 Flash beats Claude Fable 5 on Terminal-Bench 2.1, while being 11x cheaper 💰

As open-source models become more capable, they can now generate large numbers of high-quality candidate solutions and verify their own outputs at very low cost.

For example, we find that sampling just 5 solutions with DeepSeek V4 Flash and ranking them using the same model with LLM-as-a-Verifier can lead to a significant boost in accuracy (79% → 88%), outperforming closed frontier models on Terminal-Bench.

Try it out today: https://t.co/UhudjKDaYI

More on verification scaling in my previous post.

[CuiMao @CuiMao]

大家好，紧急更新CuIMao版的《牛来》，由Seedance2.5制作。