Akarui Satoshi

Efficient Procrastination

[Eric Fish @ericfish85]

MORE American students are able to get these positions BECAUSE Chinese students subsidize these programs and keep the lights on. This is not something I "believe". This is what the data tells us:
- https://t.co/TLQdrxTeN2
- https://t.co/YnVWi7eJyq
- https://t.co/LuoTVguWj0

[ACLS @ACLS1919]

ACLS is pleased to share a new resource guide on open databases for #ChinaStudies as part of the @HLuceFdn-ACLS Digital Archives Mapping Project. The guide highlights online repositories that offer free access to large primary source sets: https://t.co/vrDD3NJdKQ

[Foreign Affairs @ForeignAffairs]

“The national security bucket has grown into a trough. From climate change to ransomware to personal protective equipment to critical minerals to artificial intelligence, everything is national security now.”
https://t.co/xm3r9TEZS5

[The Economist @TheEconomist]

Generative AI could radically alter the practice of law. If it can do in 20 seconds a task that would have taken a dozen associates 50 hours each, then why would big legal firms continue hiring dozens of associates? https://t.co/wb4WMcG7rY 👇

[meng shao @shao__meng]

0->1 构建 LLM @rasbt 25k✨

@LightningAI LLM 研究员 Sebastian Raschka 非常知名的书籍 《Build a Large Language Model (From Scratch)》对应的代码示例，非常推荐大家搭配学习！

利用 PyTorch 从零开始，一步一步构建训练像 ChatGPT 一样的 LLM！

项目地址：
https://t.co/rgI11WkJUy

[David Rennie @DSORennie]

After six years and 220 Chaguan columns, it’s time to leave Beijing. It has been an adventure and a privilege. My next beat will include a dose of China, happily: I’ll be writing a new weekly geopolitics column in @TheEconomist. Column number 5 for me https://t.co/smWmtYnuZL

[The Economist @TheEconomist]

Xi Jinping hopes that science and technology will help China overtake America. If America wants to retain its lead, it must rethink its strategy https://t.co/fScv8wb2Ob

[Byron Wan @Byron_Wan]

Lord Jonathan Sumption and Lord Lawrence Collins, both former UK Supreme Court judges, have resigned as non-permanent judges on the Hong Kong Court of Final Appeal this week.

Sumption, who was appointed to the court in 2019, would make a statement “in due course” about the reasons for his resignation, while Collins cited the “political situation”.

https://t.co/V1cCsjF26J

[Simon Evans @DrSimEvans]

Very eye-opening to see who's leading the way on clean energy investment – and who's leading on fossil fuels

China: Clean energy investment 3.7x fossil fuels
Europe: 5.5x more in clean energy
N America: Biggest investor in fossil fuels
Middle East: 5.4x more on fossil fuels

[Brilliant Maps @BrilliantMaps]

Read more about the growth here: https://t.co/QiO3Z0luap

[Pedro H. C. Sant'Anna @pedrohcgs]

I am now preparing some DiD lecture slides to give a talk at Amazon, and I have created this DiD Checklist.

I am sharing them here in case some of you find it useful!

[Laia Balcells @laiabalcells]

Transition at the APSR: https://t.co/QQRSo3cECf

[Global Investigative Journalism Network @gijn]

@FT @peterjandringa @inari_ta @maxharlow @samjoiner @leahyjoseph @upyorkshire @sunyu1117 @business @Bloomberg @ChloeWhiteaker @Leonardonclt @LHreports @WIRED @VersBeton @openrotterdam @EvaConstantaras @NetraNews @paldhous @yoli_martinez @grist @HoustonChron @BmtEnterprise @NaveenaSivam @compatibilism Responses and the Reshaping of the US-Japan Strategy Amidst Taiwan Strait Military Tension (GIJN-member @tw_reporter_org):

"The project took dry military data & turned it into a compelling visual narrative to highlight the dangers of a war breaking out."

https://t.co/Et1VLc1LtL

[National Center for State Courts @StateCourts]

#Courts are using advanced technologies like natural language processing to improve civil case management. This means faster, more efficient resolution of legal matters and better access to justice for everyone. Our latest report has the details.

[Andrew Ng @AndrewYNg]

I think AI agentic workflows will drive massive AI progress this year — perhaps even more than the next generation of foundation models. This is an important trend, and I urge everyone who works in AI to pay attention to it.

Today, we mostly use LLMs in zero-shot mode, prompting a model to generate final output token by token without revising its work. This is akin to asking someone to compose an essay from start to finish, typing straight through with no backspacing allowed, and expecting a high-quality result. Despite the difficulty, LLMs do amazingly well at this task!

With an agentic workflow, however, we can ask the LLM to iterate over a document many times. For example, it might take a sequence of steps such as:
- Plan an outline.
- Decide what, if any, web searches are needed to gather more information.
- Write a first draft.
- Read over the first draft to spot unjustified arguments or extraneous information.
- Revise the draft taking into account any weaknesses spotted.
- And so on.

This iterative process is critical for most human writers to write good text. With AI, such an iterative workflow yields much better results than writing in a single pass.

Devin’s splashy demo recently received a lot of social media buzz. My team has been closely following the evolution of AI that writes code. We analyzed results from a number of research teams, focusing on an algorithm’s ability to do well on the widely used HumanEval coding benchmark. You can see our findings in the diagram below.

GPT-3.5 (zero shot) was 48.1% correct. GPT-4 (zero shot) does better at 67.0%. However, the improvement from GPT-3.5 to GPT-4 is dwarfed by incorporating an iterative agent workflow. Indeed, wrapped in an agent loop, GPT-3.5 achieves up to 95.1%.

Open source agent tools and the academic literature on agents are proliferating, making this an exciting time but also a confusing one. To help put this work into perspective, I’d like to share a framework for categorizing design patterns for building agents. My team AI Fund is successfully using these patterns in many applications, and I hope you find them useful.

- Reflection: The LLM examines its own work to come up with ways to improve it.
- Tool use: The LLM is given tools such as web search, code execution, or any other function to help it gather information, take action, or process data.
- Planning: The LLM comes up with, and executes, a multistep plan to achieve a goal (for example, writing an outline for an essay, then doing online research, then writing a draft, and so on).
- Multi-agent collaboration: More than one AI agent work together, splitting up tasks and discussing and debating ideas, to come up with better solutions than a single agent would.

I’ll elaborate on these design patterns and offer suggested readings for each next week.

[Original text: https://t.co/y4McIAjD2m]

[Hong Kong Democracy Council @hkdc_us]

The #HongKong government’s begun the process of enacting a new national security law, on top of the one imposed by the CCP in 2020. HK faces no external threats; this law, like the CCP’s NSL, targets HK people as threats to the authoritarian regime’s power
https://t.co/X2qjLaktIJ

[Akarui Satoshi @akaruisatoshi]

The Missed Chance for Peace https://t.co/AbXAaAoLkP

[Prof Helena Legido-Quigley @legidoquigley]

Our paper with the experiences of brave Spanish women in academia and the Spanish Health System “Enough is Enough”: tackling sexism, sexual harassment, and power abuse in Spain's academia and healthcare sector - The Lancet Regional Health – Europe https://t.co/BGGRxZF19w

[Guillermo Toral @guillermo_toral]

🚨 We are hiring in 3 junior lines! Come join us at @iespega, a growing community of political scientists doing exciting work in PE, CP, and IR in Madrid, Spain 🇪🇸🌞
PE: https://t.co/2aEBee96or
CP: https://t.co/Zqlsd9cU9Y
IR: https://t.co/pm9YYqDWNZ
Deadline Oct 20

[Akarui Satoshi @akaruisatoshi]

Bye, Texas. Texas Republicans Want to Reform Higher Ed. What Are Their Plans? https://t.co/XkANPgSUxV