Intro

In this episode, we explore the Chinese AI ecosystem with 'L-squared,' an anonymous tech worker based in Beijing. We discuss major players, model quality, public engagement, regulation, and the US 'chip ban.' Discover the similarities and differences between US and Chinese AI landscapes, and gain a nuanced perspective on the current state of AI in China.

USEFUL RESOURCES:

Testing Chinese models: Yi-34B-Chat (made by Kai-Fu Lee's team 01.AI) can be tried out via Replicate (https://replicate.com/01-ai/yi-34b-chat) or Hugging Face. You can also use the ChatGLM playground (https://open.bigmodel.cn/trialcenter) and Baidu's ERNIE (https://yiyan.baidu.com/) without a Chinese SIM card.

Benchmarking models: SuperCLUE is one of the most prominent benchmarks - the latest results are on GitHub (https://github.com/CLUEbenchmark/SuperCLUE) and the paper explaining the methodology is here (https://arxiv.org/abs/2307.15020).

Regulation: Explainer (https://carnegieendowment.org/2023/07/10/china-s-ai-regulations-and-how-they-get-made-pub-90117) from Matt Sheehan; piece (https://www.chinatalk.media/p/how-tight-ai-regs-hurt-chinese-firms) on how genAI regs are affecting Chinese companies.

US-China competition: Jeff Ding's work (https://www.tandfonline.com/doi/full/10.1080/09692290.2023.2173633) on the diffusion deficit in S&T; Bloomberg piece (https://www.bloomberg.com/graphics/2023-china-huawei-semiconductor/) on Huawei's semiconductor development efforts.

Staying up to date: Sign up to alerts from CSET's Scout tool (https://scout.eto.tech/); subscribe to Concordia AI's AI safety (https://aisafetychina.substack.com/) in China newsletter (disclaimer: I used to work at Concordia!)

A 2016 profile (https://chinai.substack.com/p/chinai-37-happy-20th-anniversary) on Microsoft Research Asia by Wang Jingjing, covered in Jeff Ding's ChinAI newsletter

SPONSORS:

Oracle Cloud Infrastructure (OCI) is a single platform for your infrastructure, database, application development, and AI needs. OCI has four to eight times the bandwidth of other clouds; offers one consistent price, instead of...does data better than Oracle. If you want to do more and spend less, take a free test drive of OCI at https://oracle.com/cognitive

Omneky is an omnichannel creative generation platform that lets you launch hundreds of thousands of ad iterations that actually work customized across all platforms, with a click of a button. Omneky combines generative AI and real-time advertising data. Mention "Cog Rev" for 10% off http://www.omneky.com/

The Brave search API can be used to assemble a data set to train your AI models and help with retrieval augmentation at the time of inference. All while remaining affordable with developer first pricing, integrating the Brave search API into your workflow translates to more ethical data sourcing and more human representative data sets. Try the Brave search API for free for up to 2000 queries per month at https://bit.ly/BraveTCR

Head to Squad to access global engineering without the headache and at a fraction of the cost: head to choosesquad.com and mention “Turpentine” to skip the waitlist.

Plumb is a no-code AI app builder designed for product teams who care about quality and speed. What is taking you weeks to hand-code today can be done confidently in hours. Check out https://bit.ly/PlumbTCR for early access.

TIMESTAMPS:

(00:00) Introduction

(07:24) China's AI Ecosystem

(13:40) Public AI Engagement

(17:33) Sponsors : OCI / Omneky

(18:50) AI Tools Comparison

(35:37) Sponsors : Brave / Squad / Plumb

(39:14) AI Regulatory Maze

(51:02) AI Performance, Censorship

(55:28) Chinese AI Regulations

(01:04:37) Tech, Research Role

(01:12:11) Global AI Ecosystem

(01:23:22) Cultural AI Perspectives

(01:29:14) AI Safety, Cooperation