Collective Intelligence, not AI. ⊗ Seven exponential policies ⊗ Getting materials out of the lab

‘Artificial Intelligence’? No, Collective Intelligence.

Ezra Klein interviews Holly Herndon for his podcast (full transcript at the link too). Excellent discussion on art/music and AI, Herndon’s various projects using the technology, some of her background, and the recent community/industry projects she’s been spearheading with her partner Mat Dryhurst.

A couple of weeks ago, I linked to Zoë Schlanger’s piece about plant ‘intelligence’ and finished with “in an alternative history, we progressed much further in our knowledge of fauna, flora, fungi, and their intelligence before wiping out so much of them, and AI might not have been named AI and not been created/perceived/received in quite the same way.” Herndon’s thinking made me think of that piece. Though it’s nowhere as big a leap as my fabulation, her background with church choirs, techno in Berlin, and working with AI all connects through the idea of collective work in a very though-provoking manner, quite at odds with just seeing the field as data and code. In a way, to her data and model are more akin to voices and the music.

Which is the crux of why I’m sharing it here. Herndon talks about “collective intelligence,” a commons being brought together through the model and code. Listening to her it felt like that reframing is essential, going from ‘artificial’ to ‘collective,’ towards something closer to a commons. Collective data and results made possible by private and public sector products. Something like private and public hydro projects producing electricity from a commons, the river. Or olden days sheep herders using common land but raising their own sheep.

I’m already running a bit long for one article but rediscovering/understanding/accepting our entanglement with nature, and then seeing all of what we use as a commons to be properly stewarded is essential to all our issues, it should also stretch to our data and how it’s used.

Because I really do see it as a kind of aggregate human intelligence. It’s trained on all of us. Specifically, when you look at music, it’s trained on human bodies performing very special tasks. And I think it does humans a great disservice to try to remove that from the equation. […]

And it does seem very different to me if you understand these as on some level, a societal output, something that’s built on a kind of commons as opposed to a tremendous leap and feat of technology that is the sort of individual result of software geniuses working in garages and office parks somewhere. […]

The machine is ingesting that media, if you want to call it looking at, reading, listening to, learning from. So I kind of land in that — I like to call it the “sexy middle ground” between people who are all for open use for everything and people who want to have really strict I.P. lockdown. […]

We are recording choirs across the U.K. [and] they’re joining a data trust. And we’ve hired a data trustee to pilot this idea of governance where we’re trying to work out some of the messy issues around how a data trust might work. And then we’ll negotiate with that data trust directly as to how we can use their data in the exhibition and moving forward. […]

I think what’s really needed is that people have the ability to create whatever subcultures and whatever kind of economic models work for their subcultures and aren’t squeezed into a kind of sausage factory where everything has to follow the same logic.

Seven exponential policies for a new government

Elections are coming to the UK and Azeem Azhar proposes seven polities because “the government we choose will navigate the critical years of the exponential transition: that short period until 2029.” As he mentions, most of these are actually good for a number (all?) countries. He’s further on the techno-optimist axis then I am, though pretty reasonably so. I still feel like he’s too convinced by some of his views on what’s coming within how many years. Have a read, tone it down about 20% and add MOAR climate impacts to your reflections ;-).

Launch a nationwide process to engage, deliberate and consult on key issues of technology change, from AI to energy transition. Such citizens’ assemblies can help identify local fears, needs and priorities, while building broader public trust in the system. […]

Reinforce the importance of the independence of thought that is the hallmark of science. Access and participation don’t attack the academy. […]

We know that the economy is changing and where we need to go. We should be building towards an electrified economy, a decarbonised grid, novel industrial processes, heavy use of AI and computation in the economy to support all industries and functions.

Getting materials out of the lab

At Works in Progress, good read on new materials and the hard road from research to development at scale until they reach markets and viability. In the piece above, Azhar above mentions learning rates, how technologies start expensive and get cheaper. He often mentions that we don’t tend to properly understand the speed of transitions when the learning rates accelerate. This piece shows how going from the initial discovery to benefitting from said rates can be arduous. I’d also encourage you to keep it in mind or get back to it when you see some new magical material announced. I love hearing about them, but it can be a good long while before they change anything.

Leaving the lab requires bridging many orders of magnitude: from producing less than 0.001 kilograms (one gram) per day in a lab to more than 1,000 kilograms (one tonne) per day in a factory.

§ Nice table here by Johannes Klingebiel, comparing the fantasy of technology to the real world of technology (Ursula M. Franklin). Like innovation which is fantasised to be “investment-driven, shiny, lab-born, experimental, exciting,” but has to be “needs-driven, scrappy, on location, iterative, practical, mundane.” It reminded me of Ruha Benjamin’s exercice last week at C2 Montréal, where she had the crowd emulating fantasy tech, chanting enthusiasm for a future on Mars and a dismissive “get real” for healthcare for all, to show us the limits put on our imagination. Her excellent TED Talk (yes, there are still some great ones), Is Technology Our Savior — or Our Slayer? goes in some of the same directions, emphasizing the need for UStopias.

§ I haven’t read any of the reviews yet, just breezed through a couple of blog posts, but I’m really looking forward to following the progress of Daylight, “a More caring computer.” A tablet with 60fps e-ink display, amber backlight for night reading, and a unique approach to the OS. Short doc and interview with the founder.

Algorithms, Automations & Augmentations

EU commission establishes AI Office
“The AI Office aims at enabling the future development, deployment and use of AI in a way that fosters societal and economic benefits and innovation, while mitigating risks. The Office will play a key role in the implementation of the AI Act, especially in relation to general-purpose AI models. It will also work to foster research and innovation in trustworthy AI and position the EU as a leader in international discussions.”

A sociotechnical approach to AI policy
“Generally, a sociotechnical perspective means viewing society and technology together as one coherent system. In other words, it is not possible to understand the ‘social’ without the ‘technical,’ nor the ‘technical’ without the ‘social.’ Explaining the outcomes of any technology requires focusing on the messier ‘middle ground’ between these two realms.”

Newspaper conglomerate Gannett is adding AI-generated summaries to the top of its articles
“The addition of AI-generated summaries comes as members of a local union raise concerns around proposed contract language regarding the use of AI. According to Digiday, unionized workers at the Democrat & Chronicle in Rochester, New York, were alarmed to see a clause added to their contract during bargaining that read: ‘Artificial intelligence (AI) may be used to generate news content.’”

Sony Music Group warns more than 700 companies against using its content to train AI
The group, “which owns such well-known labels as Columbia Records, RCA Records and Epic Records — has started sending formal letters to more than 700 generative AI companies and streaming platforms prohibiting them from mining text or data, web scraping or otherwise using any SMG content without explicit licensing agreements.”

Asides

• 💦 🇺🇸 🤯 East Coast has a giant offshore freshwater aquifer—how did it get there?. “Freshwater is perhaps the world’s most essential resource, but climate change is enhancing its scarcity. An unexpected source may have the potential to provide some relief: offshore aquifers, giant undersea bodies of rock or sediment that hold and transport freshwater. But researchers don’t know how the water gets there, a question that needs to be resolved if we want to understand how to manage the water stored in them.”
• 👏🏼 🧂 🔥 🇦🇺 Simple desalination tech needs just a dash of heat. “Researchers in Australia have now devised a simple new desalination method that does not need high pressure, high temperature or complex materials such as membranes. And it only uses only a fifth of the energy of conventional methods.”
• 😲 💸 🌏 Older workers pivot to gig work amid global economic crises. “Over the past three months, Rest of World spoke to 52 gig workers between the ages of 50 and 75 years in Latin America, Africa, South Asia, and Southeast Asia. Some chose gig work to keep up with rising living costs or to make up for threadbare social security systems; others say it’s impossible to find employment elsewhere once they near retirement age.”
• 🔋 ⚡️ 🚗 EV batteries old & new, could be a solution to grid storage.. “A new study suggests vehicle-to-grid technology and reused old EV batteries could meet all of the EU’s need for battery storage—and then some.”
• 😡 🥵 ⌛️ Summer 2023 was northern hemisphere’s hottest for 2,000 years, tree rings show. “Across this vast area of land, encompassing Europe, Asia and North America, surface air temperatures were more than 2°C higher in June, July and August 2023 than the average summer temperature between AD1 and 1890, as reconstructed from tree ring records.”
• 👏🏼 👏🏼 🇨🇳 Analysis: Monthly drop hints that China’s CO2 emissions may have peaked in 2023. “China’s carbon dioxide (CO2) emissions fell by 3% in March 2024, ending a 14-month surge that began when the economy reopened after the nation’s “zero-Covid” controls were lifted in December 2022.” (Via Fix The News.)