Georgios TRI

Curated content on science, technology, society, and environment.

[Scientific American @sciam]

The speedy machine displaces the U.S.’s Lawrence Livermore National Laboratory’s El Capitan at the top of the TOP500 rankings of the world’s fastest supercomputers
https://t.co/RAWbmCHqve

[Georgios TRI @yiotri]

Concerning the Spiritual in #Art by Wassily #Kandinsky https://t.co/gajV7leaR0

[Georgios TRI @yiotri]

on the method of #science (“#Physics and #Reality", #Einstein, 1936) https://t.co/NUIGnzDf7T

[Georgios TRI @yiotri]

on childlike #creativity (The Human Zoo (1969), Desmond Morris) https://t.co/HdnP7zBvxc

[Georgios TRI @yiotri]

the automatic machine https://t.co/RflY921G4Q

[Georgios TRI @yiotri]

"The machine plays no favorites as between
overall labor and white collar labor." --Norbert Wiener, The Human Use of Human Beings

[MIT Stone Center on Inequality & Shaping Work @MITshapingwork]

Why isn't the market pursuing pro-worker AI? Co-director @DAcemogluMIT explains that the reasons are both economic and ideological.

[The Economist @TheEconomist]

AI labs are well placed to take advantage of one of the big vulnerabilities of software-as-a-service companies: their siloed nature https://t.co/QhwBzaKlom

Illustration: Mike Haddad

[Georgios TRI @yiotri]

https://t.co/8P1b0DURbr

[Kathimerini English Edition @ekathimerini]

Why is the new research council failing?
https://t.co/LnHcZEM54B

[The Economist @TheEconomist]

Our analysis of Europe’s hard right shows that they are disrupting politics across the region, leaving the question of how other parties should respond https://t.co/OFyd8hciKR

[UNESCO @UNESCO]

"The future of AI depends on the choices we make as individuals, regulators, and society."

In a world of misconceptions, we need voices like @DAcemogluMIT to guide us toward a more nuanced understanding of technology's role in shaping our future.

[The Nobel Prize @NobelPrize]

BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2024 #NobelPrize in Chemistry with one half to David Baker “for computational protein design” and the other half jointly to Demis Hassabis and John M. Jumper “for protein structure prediction.”

[The Nobel Prize @NobelPrize]

BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2024 #NobelPrize in Physics to John J. Hopfield and Geoffrey E. Hinton “for foundational discoveries and inventions that enable machine learning with artificial neural networks.”

[Daron Acemoglu @DAcemogluMIT]

An important report from a team of economist led by Mario Draghi
EU competitiveness: Looking ahead - European Commission https://t.co/Cs3DSdGIqf

[The New York Times @nytimes]

The scientist David Keith has an idea to slow global warming. Opponents fear it could result in unpredictable calamities.
https://t.co/9hL8MuCLh0

[MIT Stone Center on Inequality & Shaping Work @MITshapingwork]

Will AI push workers into low-wage, low-skill jobs and further increase inequality? Or could it instead revitalize middle-skill work?

A recent @HarvardBiz article highlights research from @DAcemogluMIT and @davidautor to interrogate these questions.

https://t.co/lIMXApZRH3

[Stanford HAI @StanfordHAI]

📢 The #AIIndex2024 is now live! This year’s report presents new estimates on AI training costs, a thorough analysis of the responsible AI landscape, and a new chapter about AI's impact on medicine and scientific discovery. Read the full report here: https://t.co/NHWDCyuzm3

[IAEA - International Atomic Energy Agency ⚛️ @iaeaorg]

Did you know that nuclear power is the 2nd largest source of low carbon energy? Combining nuclear and renewable energy, we can achieve a cost-effective energy transition. #Atoms4ClimatDid you know that nuclear power is the 2nd largest source of low carbon energy? Combining nuclear and renewable energy, we can achieve a cost-effective energy transition. #Atoms4ClimatDid you know that nuclear power is the 2nd largest source of low carbon energy? Combining nuclear and renewable energy, we can achieve a cost-effective energy transition. #Atoms4ClimatDid you know that nuclear power is the 2nd largest source of low carbon energy? Combining nuclear and renewable energy, we can achieve a cost-effective energy transition. #Atoms4ClimatDid you know that nuclear power is the 2nd largest source of low carbon energy? Combining nuclear and renewable energy, we can achieve a cost-effective energy transition. #Atoms4ClimatDid you know that nuclear power is the 2nd largest source of low carbon energy? Combining nuclear and renewable energy, we can achieve a cost-effective energy transition. #Atoms4Climate

[OECD Business and Finance @OECD_BizFin]

Our Global Corporate Sustainability Report shows that institutional investors hold the largest equity portion (41%) in the 100 listed companies with the highest disclosed GHG emissions. The public sector is an important shareholder, with 18% of the shares: https://t.co/OiYXStTgvv

[Yann LeCun @ylecun]

* Language is low bandwidth: less than 12 bytes/second. A person can read 270 words/minutes, or 4.5 words/second, which is 12 bytes/s (assuming 2 bytes per token and 0.75 words per token). A modern LLM is typically trained with 1x10^13 two-byte tokens, which is 2x10^13 bytes. This would take about 100,000 years for a person to read (at 12 hours a day).

* Vision is much higher bandwidth: about 20MB/s. Each of the two optical nerves has 1 million nerve fibers, each carrying about 10 bytes per second. A 4 year-old child has been awake a total 16,000 hours, which translates into 1x10^15 bytes.

In other words:
- The data bandwidth of visual perception is roughly 16 million times higher than the data bandwidth of written (or spoken) language.
- In a mere 4 years, a child has seen 50 times more data than the biggest LLMs trained on all the text publicly available on the internet.

This tells us three things:
1. Yes, text is redundant, and visual signals in the optical nerves are even more redundant (despite being 100x compressed versions of the photoreceptor outputs in the retina). But redundancy in data is *precisely* what we need for Self-Supervised Learning to capture the structure of the data. The more redundancy, the better for SSL.
2. Most of human knowledge (and almost all of animal knowledge) comes from our sensory experience of the physical world. Language is the icing on the cake. We need the cake to support the icing.
3. There is *absolutely no way in hell* we will ever reach human-level AI without getting machines to learn from high-bandwidth sensory inputs, such as vision.

Yes, humans can get smart without vision, even pretty smart without vision and audition. But not without touch. Touch is pretty high bandwidth, too.