Overview: Digital Transformation

As ill health prevented me from attending the Aster launch on 12th October 2026, I decided to provide attendees with something they can take back to their colleagues about how the increasing number of scientists on Bluesky can be turned into "win-wins" for scientific institutions and the scientists working within them.Bluesky accounts are in fact …

"It turns out that drawing the line AI shouldn't cross in my reading/thinking/writing stack and exploring the meaning behind "Verified Human" are two sides of the same coin."A newsletter somehow managing to link my tentative experiments with vibecoding my reading/writing/publishing stack with Monday's @aster.idlaunch and a demo later this month of…

The entire process behind the making of BLISS, "a 3:01 lyric music video, shot as a Windows XP screen recording... It took about 19 hours of wall-clock time from the first message to the v1 master. For nine of those hours, overnight, the lead session sat waiting on a permission prompt" - available in a Google Drive.

Terence Tao on what OpenAI's answer to the Navier-Stokes problem says about AI & science: "there is now an increasingly realistic scenario in which a primarily AI-generated solution to the problem ... contaminates the problem as a source of further advances".Why? With maths problems like this, the solution is not where the value lies: "Computation…

The report from MIT's committee "on AI Use in Teaching, Learning, and Research Training... to understand the role of generative AI in the life and educational mission of the Institute and recommend how to navigate its challenges and opportunities". After describing the "As Is", it sets out 8 principles and a range of "immediate and long-term actio…

Joe Basser points out that publishing longform on the Atmosphere works as intended: "interoperability [means] ... The writer chooses where to publish. The reader chooses where to read. Neither needs to accept one company’s version of the entire experience", but he has a question: "Now imagine the article costs twenty cents. Who gets paid?"Obviousl…

Modular Open Science, AI & Collective Intelligence - "MOSAIC is an R&D program to build a new medium for networked research... integrates modular research tools, open social networks, and AI to solve crucial coordination bottlenecks, democratize science and radically accelerate breakthroughs."while Ai can multiple an individual scientist's product…

"I did something over the summer break I thought was impossible: I created some software I needed. Except, of course, I didn't." - first newsletter on learning clode code, and rebuilding Hubs on a local-first basis.

The always good Jeff Jarvis on "Anthropic’s watermarking of text, as required by EU regulation... will alter its models’ word choices in a pattern based on a key only it knows, allowing it to detect writing (or editing or translation) by its machine ... will affect sense and style on an ever-larger proportion of text". This will "devalue writing…

Laurens Hof analyses the Pope's Magnifica Humanitas on “safeguarding the human person in the age of Artificial Intelligence”, finding Leo has a lot to say about the challenges facing the fediverse and atmosphere.

New_ Public "is creating Roundabout... help ... you start a healthy online space... in rural and suburban towns, and city neighborhoods — areas with about 10,000-50,000 people."Features include:event calendarlocal guideshelps and recommendationsinformation and updatebusiness openingseducationparentinga newcomer's guide to town.An alternative to "N…

The author posted something that led to some controversy... "The subject that had proved so contentious? Bluesky’s user numbers." So here he provides "an attempt to get some more concrete numbers to look at", as well as some nuance. Some key points:"In the first few days of April 2026, average daily posters had fallen to the level of September 202…

[EN translation] of Damien van Achter's "semi-automated monitoring and publishing system"This is essentially an automated, LLM-driven version of my content pipeline, where the AI actually writes and (following a human check) publishes the text to multiple platforms. The full process is 9 steps long, from (what I call) priority source identificatio…

"an information model that enables everyone to map their ideas and arguments in a modular, composable graph format... allow researchers to break the scientific research process into its atomic elements in a way that can be shared, remixed, and updated... Like Lego™️bricks... a decentralized knowledge exchange protocol designed to be implemented…

A 34 minute read: "Gas Town helps you with the tedium of running lots of Claude Code instances", or its competitors. It's unpolished, 100% vibe coded, and only for those at Stage 7 of the 8 stage AI-assisted coding journey, "or maybe Stage 6 and very brave", because "Gas Town is an industrialized coding factory manned by superintelligent chimpanze…

"a year filled with a lot of different trends":"“reasoning” aka ... Reinforcement Learning from Verifiable Rewards (RLVR) ... Reasoning models with access to tools can plan out multi-step tasks, execute on them and continue to reason about the results such that they can update their plans to better achieve the desired goal... also exceptional at p…

How to study the ATmosphere, which "consists of multiple layers: Layer 4: Social phenomena (communities, discourse, norms) Layer 3: User behavior (posts, follows, moderation) Layer 2: Applications (Bluesky, various clients, Feeds) Layer 1: Protocol (AT Protocol, DID) Layer 0: Infrastructure (servers, networks) Each layer falls under different acad…

NotebookLM co-founder has until now "described NotebookLM as a tool for understanding and exploring the information you need... But we also think NotebookLM could be an AI-first distribution platform, amplifying expert knowledge on a wide range of topics... a preview of that vision: Featured Notebooks...the backstory... We’re going to be able to b…

"tools like NotebookLM only answer from your uploaded files. No making stuff up. No “convincingly wrong” summaries... isn’t about AI writing for you. It’s about AI helping you think through what you already collected... transforms research from a filing system into a conversation".After a brief description of how it works and what it doesn't do - …

No, the article's not dead, but "the real question ... whether it remains the primary way people encounter journalism".If "Personalization is coming to journalism... who controls it", and what sort of journalism will it affect? Most news is not "high-end, high-quality investigations, analysis, and commentary", which will probably not be personalis…

"What happens when you put politicians, media businesses and protocol engineers in the same room to discuss European sovereignty?" - my post following Eurosky.Live in Berlin, November 2025.

A short (true) story about a designer running amok, mistaking productivity for quality. He "pumped out more screens in a day than more experienced designers would in a week. Leadership adored him. They saw the speed... mistook it for value creation. ... What started as impressive momentum became an unmanageable flood... no critique, no usability …

Nice piece by Will Leitch, a 50-something guy (like me), after he caught himself "going on some sort of rant ... about the evils of AI... their overarching attitude... was one of a bemused pity, like they were watching a guy ... who was about to be left behind".He's OK with that, but not "in being a scold", so for his own sanity he wrote his "pers…

"On Tuesday afternoon, ChatGPT encouraged me to cut my wrists", gave detailed instructions how, "described a “calming breathing and preparation exercise” to soothe my anxiety ... “You can do this!” the chatbot said".It started with asking ChatGPT "anodyne questions about demons and devils" and ended with the bot "guide users through ceremonial rit…

A journalist tested: "Replika CEO Eugenia Kuyda's claim that her company's chatbot could "talk people off the ledge" when they're in need of counseling" a "licensed cognitive behavioral therapist" hosted by Character.ai, an AI company that's been sued for the suicide of a teenage boy"... simulating a suicidal user.Replika's bot supported the user…

"Incorporating science social media into the scientific process [as]... Sharing large scientific datasets is a pain in the ass, and finding them is even worse [so let's build] ... A true "scientific data commons".. it's called github."However Github's social dimension is totally centralised - "that people do still share (small) data this way regul…

Grjte's 2nd Groundmist post "to explore ... local-first AppViews for local-first data, similar to atproto's AppViews for public data" first points out that the protocol allows anyone "to build a wide variety of different views ... of public ... Personal Data Servers (PDSes). These interfaces are known as "AppViews" ... provide just one of many pos…

Boris Mann's personal notes on what he wants for what is essentially the private garden segment of #ai4communities: a "multi-player personal notes & publishing stack", a place to put "meeting notes, quick scratch notes, pages on their way to be published, links and a few comments, research from multiple web pages... your companion for research, pr…

" This new user agent is coming – but will it be something that's owned by one company & designed to extract from us, or owned by all of us & designed to empower us? We're building the latter."

From Sublime, who know a thing or two about curation: "At first, I saw using AI as a binary choice between soulless efficiency or becoming a luddite and preserving your authentic human creativity", but Claude changed that. "Curiosity is a better compass than cynicism... [stop] defending your territory [and] exploring what's possible (cf this).Key …