Commonplace book: supply chains, hope, cool
Yes this:
 |
| https://social.coop/@luis_in_brief/115696408257250551 |
Helen Toner on generative AI and where it might go, which is a rare article on foundational models and the real world. Hard to pick highlights, but here goes:
Two things are true about AI, I claim:
AI models keep getting better and better
AI models keep sucking, and the things they keep sucking at are kind of confusing
... So the bold claim in this talk is: maybe AI will keep getting better and maybe AI will keep sucking in important ways.
... Relatedly, pushing AI-for-good forward will matter: I didn’t say this in the talk, but related to the previous point—jaggedness raises the stakes for anyone hoping to use good AI to counterbalance against bad AI, so to speak. Two major examples of this are using AI to automate alignment or other safety research, and using AI for societal resilience measures like biodefense or cyberdefense. If AI progress is not jagged, it might make sense to just hold off and wait for AI to get better before trying to push hard on these. If progress is jagged, then it’s important to push these forward as hard as we can now, in the hope that they end up being one of the more advanced pointy parts of the jagged frontier, not one of the lagging hollow parts.
... We should be thinking more about human-computer interaction dynamics, human factors, user experience, user interface. What does the trust between those parties look like? How do you maximize the benefit that you’re able to get from those AI systems by designing that human-AI system well? As opposed to just skipping to the end and assuming, well, at some point the AI is going to automate all of it, so who cares about the human-computer part.
... For a long time, I have not found a very helpful to talk about timelines to AGI, because I think people use just such different metrics and definitions for what they mean by that. I tend to think more in terms of how crazy are things going to get, how soon? And you can use different metrics for crazy, and I find that maybe a more productive and more interesting way to think about things.
Ryan Broderick has some interesting observations[paywall, worth it] about the internet and Gen Z:
And I think offline shock has a strong connection to aesthetic disappointment: “Why doesn’t my life look as good as this image I found online?”....
What ties these two online meltdowns together is the breakdown of the online life and the offline one. In the case of the cozy living room, it’s the extremely uncomfortable revelation that their less online peers have friends (lol sorry). Or housing, I guess? And in case of the Pitchfork lists, it’s the idea that all the metrics X stans use to define their favorite artists — views, sales, streams — can’t make them actually cool. Not “popular,” but cool. Both of these little online misadventures are about the crushing realization that it’s very hard to be authentic online right now. And this is why I think when Gen Z makes traditional art — film, music, etc. — it can come off as cosplay. The feeling like all of the options they could possibly make have been mapped out for them already. Like Lego blocks they can piece together as they please. And it’s why I’d argue the defining art form of Gen Z is incomprehensible brainrot. Which I do consider art and, incidentally, is the most authentic thing I’ve seen young people make.
Martha Lane-Fox notes "we have created a public sphere that deters exactly the kind of thoughtful, capable, values-driven people we need to lead it."
We are living in a culture where accountability has morphed into instant annihilation. Yes, leaders must answer for their mistakes. But the speed at which scrutiny now turns into spectacle is breathtaking. Westminster is the most extreme example: five prime ministers in six years, 16 housing ministers in a decade, entire departments resetting their leadership teams almost annually.
...Trust is now at structurally low levels. ... In that environment, even good policy struggles to work. Too few people believe in institutions strongly enough to give them the benefit of the doubt. Every decision becomes fraught. Every reform gets bogged down. Cynicism quietly becomes the national operating system.
A deep dive into the COVID Inquiry in the UK and how it is not helping us get ready for future crises. HT Rachel Coldicutt.
A core issue is that the inquiry is deeply flawed in ways that risk being counter‑productive — ie, it risks being anti-useful. It is actively drawing demonstrably wrong conclusions, including making strong conclusions on issues that are ultimately political value judgments best assessed by the public, as well as drawing strong conclusions on issues that one will never be able to have confidence on. It is blocking the space that should be occupied by the vitally needed effective process to get us ready for next time.
... The system’s incentives appear aligned more with internal process preservation than with the rapid learning required after a national crisis.
... The core issue is that the inquiry is itself a striking manifestation of the very problems it should be trying to address, whilst simultaneously being deeply unclear about what it is actually trying to address.
... It goes to the heart of the matter: a late-Victorian, hierarchy-obsessed model of government combined with a legal-permanent bureaucratic state laid on top is fundamentally misaligned with the complexity and speed of 21st-century crises.
I have a big backlog of links about the current state of open which I may one day synthesise. In the mean time, I came across Erlend Sogge Heggen writing:
Since its inception the free and open source software movement has lacked a theory of change beyond the liberation of computer code. Liberation of human beings by way of a liberatory technology was always a secondary and oftentimes incompatible concern for Open Source, since laborers having agency of their work (and how it may be exploited) is in conflict with the inviolable liberties of an Open Source computer program.
The result has been an ineffectual "open source revolution" that maintains the status quo of our modern day hellscape by facilitating an upwards transfer of wealth and power, amassed by the hyperscalers who are now entering their final, fascistic form. Open source "won" by aiding and abetting the already dominant owner-class.
... Anti-fascistic software..is made possible by pro-labor licensing.
... The intellectual property landscape is grossly outdated and biased towards big business, but the market still depends on the continued rule of law to function. That systemic dependency can be subverted against itself and used in our favor by reformulating acceptable use for the people's software at a grassroots level.
Ethical licensing is hard, but far from impossible. While we cannot easily encode anti-fascism into our licenses, what we can easily do is discriminate against big capital, the underlying engine of fascism and authoritarianism.
Glyn Moody knows open source. He notes an interesting potential:
More recently, many leading AI systems have been released as open source. That raises the important question of what exactly “open source” means in the context of generative AI software, which involves much more than just code. The Open Source Initiative, which drew up the original definition of open source, has extended this work with its Open Source AI Definition. It is noteworthy that the EU has explicitly recognised the special role of open source in the field of AI. In the EU’s recent Artificial Intelligence Act, open source AI systems are exempt from the potentially onerous obligation to draw up a range of documentation that is generally required.
... Article 3 of the CDSM Directive enables these [research] institutions to text and data-mine all “works or other subject matter to which they have lawful access” for scientific research purposes. ... the use of open source licensing is critical to this application of Article 3 of EU copyright legislation for the purpose of AI research.
... What’s noteworthy here is how two different pieces of EU legislation, passed some years apart, work together to create a special category of open source AI systems that avoid most of the legal problems of training AI systems on copyright materials, as well as the bureaucratic overhead imposed by the EU AI Act on commercial systems.
... public AI systems provide a way for the EU to compete with both US and Chinese AI companies – by not competing with them... But public AI systems, which are fully open source, and which take advantage of the EU right of research institutions to carry out text and data mining, offer a uniquely European take on generative AI that might even make such systems acceptable to those who worry about how they are built, and how they are used.
We all need hope.
I went to one of the Society for Hopeful Technologists events in November, and we worked on the draft Charter. Looking forward to seeing how this develops in 2026!
Bonfire ("Public Interest Social Networks") did a crowdfunder. I love that they focus on community first, and the tools communities need to get together and do stuff; I also love that are working within a network of different projects, not just building yet another thing. One to watch.
Jack McGovan writes about a project to bring public diners/canteens to Scotland - what if dinner was public infrastructure?:
While charity or pay-what-you-can models offer a way for the needy to eat, they’re a symptom of a failure in social policy. That’s why the food charity Nourish Scotland are pushing for the reintroduction of public diners to Scotland: diners subsidised by the state to provide affordable, nutritious, and filling meals that are accessible to everyone regardless of their financial status.
Alex Deschamps-Sonsino writes about how design is not going to save us - because it hasn't helped much so far. (She's also looking for work, and will be an amazing thoughtful design leader for someone!)
I'm not sure I'll be able to go, but I am excited that there's a community and a conference around rewilding gardens to support nature. Rewilding agricultural land is not always a great idea, but gardens are different.