Anthony Howell

Director of Center on Tech, Data, Society and Professor of Public Policy @ASUPubAffairs. Past Econ Prof @PKU1898. GEO PhD @UCLA. Path less taken.

[Joshua Meier @joshim5]

We’re excited to introduce @ChaiDiscovery and release Chai-1, a foundation model for molecular structure prediction that performs at the state-of-the-art across a variety of drug discovery tasks

We're releasing inference code, weights & a web interface: https://t.co/QmpbVO9Fhd

[Andrej Karpathy @karpathy]

The hottest new programming language is English

[Jennifer Doleac @jenniferdoleac]

Super smart play by @ssrc_org. If your university admin is a nightmare, you can run your grants through them instead!

#competition

https://t.co/RGhdrfnIUU

[Li Zexin 李泽欣 @XH_Lee23]

There's a little-known story behind these beautiful tunnel-boring machines (TBMs):

🔻In 1997, when China was building the Qinling Tunnel, we paid RMB 700 million for 2 SECONDHAND machines from Germany. At that time, this amount was equivalent to the total tax revenue of an entire region. This was daylight robbery when China couldn't produce the machines itself.

Additionally, the RMB 700 million did not include any technical support or after-sales service. We also had to spend 800 euros a day (around RMB 6,800) to hire German engineers for maintenance. At that time, the average annual salary of a Chinese worker was less than RMB 4,000.

During that period, the global TBM market was monopolized by the US, Japan, and Germany.

🔻In 2002, China decided to embark on a challenging journey of catching up. The TBM is a very complex system with over 30,000 precision parts.

In 2008, after 6 years of effort, China’s first TBM with independent intellectual property rights was completed.

To prove its reliability, China conducted a comparative test during the Chengdu Metro project in 2012.

Two TBMs, one imported and one self-produced, worked simultaneously for a project. The result showed that the self-produced TBM was comparable to the best in the world.

Following this, China’s self-designed TBMs began to be manufactured and used in major projects across the country.

🔻Today, China is also exporting TBMs and has gradually gained a 70% share of the international market. Unlike before, a Chinese TBM now costs only RMB 25 to 50 million.

Meanwhile, the market share of German TBMs has sharply declined.

[Anthony Howell @AntJam_Howell]

@RobinatASU Big congrats @RobinatASU

[Stack Overflow @StackOverflow]

We’re thrilled to announce we’re partnering with @OpenAI to bring best in class technical knowledge and the world’s most popular LLM models for AI development together! This groundbreaking partnership with OpenAI will drive our mission to empower the world to develop technology through collective knowledge.

https://t.co/CbgZIEoB2z

[Andrew Lokenauth @FluentInFinance]

You can now use ChatGPT inside Microsoft Excel or Google Sheets.

Here's how:

[Dina D. Pomeranz 🟣 @DinaPomeranz]

A strong political divide is emerging between young men and women in many countries:
https://t.co/kPY21noAt8

[Luca Bertuzzi @BertuzLuca]

LEAK: Given the massive public attention to the #AIAct, I've taken the rather unprecedented decision to publish the final text. The agreed text is on the right-hand side for those unfamiliar with four-column documents. Enjoy the reading! Some context: 1/6
https://t.co/0EmhajXB2x

[Karen Mossberger @KMossberger]

Looking forward to the January 29th cohort for the online Digital Inclusion Leadership Certificate! See https://t.co/ELbUgCHrd0 to get more info and enroll! #Digitalinclusion #Broadband @MarconiSociety @ASUPubAffairs @asupublicsvc @ASU_CTDS

[NASAEarthdata @NASAEarthData]

Data from NASA's Socioeconomic Data and Applications Center reveal what Earth observations mean for the lives of people and communities on our planet. Read how socioeconomic data put the who into the scientific questions of what, where, when, why, and how: https://t.co/nLFUWxSW6O

[Michael Crow @michaelcrow]

This is one of @ASU’s most significant research projects, one that will have direct and positive impact across all communities. https://t.co/4ru9giIFBQ

[Sari Arho Havrén @SariArhoHavren]

Economy in China falls now under national security: “The ministry [of State Security] believes that excessively realistic explanations about the harsh economic situation – or, in its parlance, repeating the false theory put forward by foreign forces that China is in economic decline – threaten national security. As such, it is signaling that those who go negative in their economic analyses could be subject to punitive action.”

https://t.co/VyeRZ44Lgi

[Harsh Nisar @NisarHogaya]

🚨🛣️🛰️PMGSY has silently released GIS data for 90% of the rural roads built under the program since 2000.

5.7 lac km from PMGSY I, II & III.

Look at the scale of work done by this scheme! Such a privilege to have worked for it.

Download Link: https://t.co/qCZNlmsbXf

[QJE @QJEHarvard]

Recently accepted by #QJE, “Logs with Zeros? Some Problems and Solutions,” by Chen (@jiafengkevinc) and Roth (@jondr44): https://t.co/id0m8gz2Kr

[James Pethokoukis ⏩️⤴️ @JimPethokoukis]

Google DeepMind used a large language model to solve an unsolvable math problem https://t.co/zvEtTcuyRE

[Daniel Di Martino @DanielDiMartino]

Javier Milei's minister of economy just announced an "emergency package" of measures to completely balance the budget in 2024 equivalent to over 5% of GDP.

This would be equivalent to a $1.4 trillion austerity package in a single year in the U.S. economy.

The measures include:
- Letting go of every public employee who has been employed for less than 1 year
- Abolishing all government publicity and related spending for projects.
- Appointed government positions are cut by 34%
- No more transfers to local governments
- Suspension and cancelation of all public infrastructure (subject to a lot of corruption in the past)
- Reduction of energy and transport subsidies
- Devaluation of the peso from 350 to 800 pesos per dollar
- Elimination of export and import quotas and licenses
- Temporary increase in non-Agricultural taxes for exports and imports to the same level of the agricultural industry for uniformity
- Temporary expansion of direct aid through the child benefit and food aid debit card

https://t.co/czhD5hLDZN

[Channel 1 @channel1_ai]

See the highest quality AI footage in the world.

🤯 - Our generated anchors deliver stories that are informative, heartfelt and entertaining.

Watch the showcase episode of our upcoming news network now.

[Anthony Howell @AntJam_Howell]

@adrian_horning_ I’m interested.

[Bindu Reddy @bindureddy]

Large Language Models Represent Space and Time

It's just been a few months since Llama-2 has been open-sourced and already there are several labs doing some foundational research and enhancing our understanding our LLMs

Earlier today, Wes Gurnee and Max Tegmark published an exceptionally intriguing paper.

Fundamentally, they dive into how much LLMs really understand the data they are trained on and how much of it is just fancy statistics.

The find that LLMs are much more than "stochastic parrots" and learn linear representations of space and time

The researchers analyzed the learned representations of LLMs on three spatial datasets (encompassing world, US, and NYC places) and three temporal datasets (comprising historical figures, artworks, and news headlines) within the Llama-2 family of models​

Their analysis demonstrates that modern LLMs acquire structured knowledge about fundamental dimensions such as space and time, supporting the view that they learn not merely superficial statistics, but literal world models.

LLMs they find learn linear representations of space and time across multiple scales. These representations are robust to variations in prompts and unified across various entity types like cities and landmarks

The paper mentions the identification of individual "space neurons" and "time neurons" within Large Language Models (LLMs). These specific neurons are found to reliably encode spatial and temporal coordinates respectively, showcasing how the LLMs have structured, inherent mechanisms to represent and understand fundamental dimensions like space and time in the data they process

This paper is an interesting step towards how LLMs can learn and represent complex, real-world information in a structured and meaningful manner. While this doesn't mean that we suddenly have stumbled into AGI :), it does contribute to our understanding of why LLMs are so exceptionally performant and delightful.

A huge congrats to the authors and I am hopeful this is just the one of the many papers about this subject that will be published in the next year

The more we understand how LLMs work, the more we can dispel the fear, superstition and apprehensiveness about AI research and the faster we can make progress