@gerald_leppert@bonn.social❔ How can climate mobility function as a proactive strategy for women and marginalised groups? What influences decision-making processes to migrate or to stay? What do governments and stakeholders need to do to increase women's agency?
📝 New Paper Published:
Human mobility as a climate and disaster risk management strategy: Addressing intersectional vulnerability and enhancing women's agency
🔗
@ErikJonker@mastodon.social"Een veel kleinere begroting: met die missie gaat premier Rob Jetten naar Brussel." , dan wil je dus niet begrijpen hoe de wereld is veranderd en in welke geopolitieke context we nu zitten....
https://www.nr…
@cdarwin@c.imThe Frey Wille Hundertwasser Street Rivers collection testifies to his commitment to architecture
and is influenced, for example, by the famous Hundertwasserhaus in Vienna.
Hundertwasser was strongly committed to architecture and fought for a more natural and human-friendly construction.
He despised rational, sterile architecture and demanded more space for individual design.
In this context, he campaigned for a "tree obligation" and a "window right&…
@fanf@mendeddrum.orggood grief i thought bournegol was bad but there are still new depths to plumb
https://www.tuhs.org/cgi-bin/utree.pl?file=V7/usr/src/cmd/sh/mode.h
@aral@mastodon.ar.alDonno what’s worse, the verse itself or DuckDuckGo’s generated explanation of it. @… https://partyon.xyz/@nando161/116925119988965039
@cosmos4u@scicomm.xyz100 years ago #OTD the #planetarium in #Jena (Germany) opened to the public: it's the oldest one in the world srill operating in the same location! Scroll down on https://skyweek.wordpress.com/2026/07/18/die-leuchtenden-nachtwolken-sind-weiter-aktiv for some news coverage back then and the wider context.
@Dragofix@veganism.socialSupport local people to protect world's nature, new report urges, as deadline for global conservation target looms https://phys.org/news/2026-05-local-people-world-nature-urges.html
@darkrat@chaosfurs.socialUpcoming #Backrooms Movie used #blender for most of the #VFX :blobratheart:
@berlinbuzzwords@floss.socialScatter-read is one of the most stubborn performance bottlenecks in distributed search. Rishav Sagar and Tejas Shah are at #bbuzz26 with a practical solution: context-aware segments.
Learn more: https://2026.berlinbuzzwords.de/session/context-aware-segments-solving-the-scatter-read-problem/
@gwire@mastodon.socialVideo talking about how UK city planning works (in the context of Sim City) which pairs well with all the announcements about AI tools for planning.
(jumpscare at 13:11)
https://www.youtube.com/watch?v=5hgJ2gKE09s
@arXiv_csLG_bot@mastoxiv.pageCrosslisted article(s) found for cs.LG. https://arxiv.org/list/cs.LG/new
[7/8]:
- MedRLM: Recursive Multimodal Health Intelligence for Long-Context Clinical Reasoning, Sensor-Guid...
Aueaphum Aueawatthanaphisut
@Techmeme@techhub.socialAt AWS Summit, AWS announced Continuum, which uses AI to find and fix code vulnerabilities, Context, which organizes company data for AI agents, and more (Todd Bishop/GeekWire)
https://www.geekwire.com/2026/amazon-unvei
@netzschleuder@social.skewed.defacebook_friends: Maier Facebook friends (2014)
A small anonymized Facebook ego network, from April 2014. Nodes are Facebook profiles, and an edge exists if the two profiles are "friends" on Facebook. Metadata gives the social context for the relationship between ego and alter.
This network has 362 nodes and 1988 edges.
Tags: Social, Online, Unweighted, Metadata
@drgeraint@glasgow.socialFascinating probability analysis of English council election voter flows (2024/2026).
https://oskanberg.github.io/2026-voter-flows/
H/T @…
@Tupp_ed@mastodon.ieAs a small trial, I’m giving a joint timeline and posting app for Bluesky and mastodon a go. This means I will see both places and post in both places in the one feed. But it also means I may say a thing prompted by something I saw on one platform without context on the other. My apologies in advance for any confusion.
@aredridel@kolektiva.socialRE: https://eldritch.cafe/@miranda_blue/116935608086213327
This is showing one thing that I think should become general principle: Using an LLM for someone else _is rude and bad_. Presenting something untranslated that _they_ can have an LLM translate is better than doing it yourself, and not showing that input — the prompt here is not included and is a key piece of context, and the results are also bad _with no recourse_.
Information has been destroyed. Trust has been broken (or in the case of commerce, failed to be established). Even if the total amount of LLM-use were the same, this is worse than the reader using it.
PROMPTING AN LLM FOR SOMEONE ELSE IS RUDE.
@tschfflr@fediscience.org🤔 Are #emojis a universal language? - No, for two reasons. 1) They are not a language, as they do not show the properties that human languages have: grammatical structure and free combinatorics, enabling us to express complex thoughts. 2) They are not really universal. While many emojis are interpreted similarly all over the world (I guess it's pretty clear what one could mean by 🍕 ), some emoji meanings are closely tied to cultural assumptions (e.g., some of the gestures 👍 , buildings ⛩️ , food), and others are very context-dependent and might follow different conventions in various groups, including linguistic or regional groups (e.g., 🙃 for which we found strong differences between populations in Norway/US vs. Germany/Bulgaria). #WorldEmojiDay #linguistics
@davidaugust@mastodon.onlineYes. It is real. He actually posted it.
(Context: Blue Angels, US Navy precision flight team, flew above beachgoers in Florida and jet wash, the kinetic energy of flight, hit the beachgoers. The flyover is in violation of the law, Navy policy and basic understanding of pilot responsibility for anyone older than 5.)
The flyover: https://yout…
@michabbb@social.vivaldi.net🍝 It kills the modal spaghetti: no separate state variables, no context providers, no Portals, no Redux or Zustand boilerplate — just an async invocation like window.confirm that waits for user interaction and hands the result straight back
🪶 Tiny footprint with under 1 KB bundle size, zero dependencies, full #SSR support and
@grahamperrin@bsd.cafe@…
<https://www.reddit.com/r/truenas/comments/wfevxe/comment/oo8jg7h/?context=1> last month (in response to a 2022 comme…
@aardrian@toot.cafeI also saw that uncredited, context-free photo in my timeline.
This entire thread is a good explanation of the many ways that post was bad form:
https://mas.to/@meganL/116888106293249758
Give credit, give context, don’t do inspo porn.
@grumpybozo@toad.socialSeriously folks, if you are discussing global non-profit organizational policy in the context of Delaware law, you should do ALL of your own comprehension and analysis or SHUT THE FUCK UP WITH YOUR CLANKER BULLSHIT!
#SubToot.
@inthehands@hachyderm.ioI can image a developer parallel to the first, too: the human still using all their skills and experience, but with the machine catching mistakes, providing context and validation and vigilance that is •orthogonal to• testing and type checking and code crafting and — the big one! — actually •thinking• about the problem.
That’s a regime I imagine developers would feel a lot better about. And I know there are people out there pursuing it! But they’re not the ones dominating the conversation.
4/
@Techmeme@techhub.socialNew York-based Jedify, whose platform connects to enterprises' knowledge sources via APIs to build a "context graph" for AI agents, raised a $24M Series A (Ram Iyer/TechCrunch)
https://techcrunch.com/2026/06/10/jedi
@ErikUden@mastodon.desomeone explained to me why they had three instagram accounts and i really didn't get it. like, a private one (sure, whatever) but then a... secondary private one (was the first one for friends and the other for family?? you're not even a public figure, what does public mean in this context??)
recently i met someone with 4 instagram accounts (this person was a cis white man who didn't transition or had multiple names or anything) and i just... didn't get it. Like, wh…
@jerome@jasette.facil.servicesEditorial: The Human Rights Museum’s Nakba exhibit ignores the important context that I don’t want to hear about it
https://www.thebeaverton.com/2026/07/editorial-the-human-rights-museums-na…
@tschfflr@fediscience.orgIs emoji interpretation just a "fun" summer topic, or does emoji misinterpretation ever become serious, like in court discussions? - In face there have been several cases both in North America and Europe where emoji meanings were negotiated in court! I keep a collection of them because it's a nice intersection with my alter ego, the "forensic" linguist. For example, in cases from Canada and Bavaria, judges had to decide whether a thumbs-up 👍 in a chat was meant as a signal of agreement (to a contract or contract amendment), or whether (like the senders argued) it constituted only an acknowledgement of receipt of the message. In both cases, the context was super important - how did the parties interact before, what else was said in the chat, etc. Interesting! #emojis #linguistics #WorldEmojiDay
@cdarwin@c.imTrump landed in Beijing last night with Elon Musk, Tim Cook, Jensen Huang, and 15 other American CEOs in tow
— the most powerful collection of corporate America ever to board Air Force One
— while at home Republicans scrambled to explain away his statement that he does not think about Americans’ financial situation.
Mike Johnson, asked directly whether that was the right message, said he did not know the context in which Trump made that comment.
The context was a p…
@frankel@mastodon.top'Never use double for money' is dogma, not engineering.
Guest author Stefano Fago breaks down when double, BigDecimal, or fixed-point is the right call, and the production traps that quietly undo each one.
https://blog.frankel.ch/bigdecimal-vs-double/
@shriramk@mastodon.socialIsaac is the fan favorite, but one interviewer who deserves more notice is Mishal Husain. She does her research, asks solid questions, follows up, gives people rope, and then footnotes both for context and to expose their inconsistencies and lies. Weekly!
https://www.bloomberg.com/mishal
@hiimmrdave@hachyderm.ioany loop that allows an LLM to modify its own behavior is necessarily going to result in undesirable emergent behavior. this includes editing tests, agent/soul files, and any other context, input, and prompt.
@samvarma@fosstodon.orgWow. This is especially interesting in the context of the World War II bombing raids over Japan, where they couldn't figure out why their extremely intricate Norden analog bombsight was off. They were not taking the jetstream into account.
Referencing "The Bomber Mafia" here, which is utterly fascinating, and might be available as an audiobook from your library, which is how I consumed it.
@arXiv_statAP_bot@mastoxiv.pageContext-Aware Optimization of Follow-Up Intervals for Type 2 Diabetes Care Using Markov Decision Processes
Parisa Lotfibagha, Kristen Miller, William J. Gallagher, Elizabeth B. Selden, Muge Capan
https://arxiv.org/abs/2606.19092
@mgorny@social.treehouse.systemsHuman translators manage good #translation because they comprehend the original text, know its intricacies and can actually creatively write to preserve its gist. Context is more than the surrounding paragraphs: it includes the cultural and social background, other books, etc. Sometimes it literally involves getting hints from the author. And in the end, it is deeply personal. It can't be "accurate", because no two people are the same, and no two languages map 1:1 (though the interpretation will also differ between different users of the same language); the best you can do is add translator's notes.
LLM translators manage good translation because they've been fed tons of human translation work, and quite often they manage to match an original human translation and plagiarize it. They may even seem better because they seem to manage it without sufficient context, but that's only because they've found a human translation that had that context already.
Or, well, technically they can also regurgitate something that makes sense only in the irrelevant original context.
#NoAI #NoLLM
@Techmeme@techhub.socialSubquadratic launches with a $29M seed and debuts SubQ, an LLM that uses a subquadratic sparse attention architecture to achieve a 12M-token context window (Kyt Dotson/SiliconANGLE)
https://siliconangle.com/2026/05/05/subquadratic-launches-29…
@sean@scoat.es@… is one of the most thoughtful, understanding, and pleasant designers I've ever had the privilege of working with.
(In fact, I hope I get to work with him again on something.)
If you're looking for detail-oriented, smart design—that only a real human can bring—I fully recommend reaching out to him at Obvious Things.
@michabbb@social.vivaldi.net📊 Results: text context produced 50 successful games at ~9 minutes average. Image context produced only 23 successful games at ~20 minutes average. Damage numbers were comparable or slightly better with images, but the runtime gap was drastic.
💸 Token usage: 306 million tokens for the text runs vs 213 million for the image runs — with roughly half as many completed games.
@mcdanlj@social.makerforums.info
@markhburton@mstdn.socialLooks like NESO are trying to prevent the kind of issue (high solar input causing overvoltage) that led to Spain's grid collapse. In Spain this was in the critical context of nuclear centrales being off-line. In Britain the availability of French interconnector supply is impacted by their nuclear shutting down as cooling water too hot.
Britain’s biggest community solar farm forced to shut over grid overload fears | Solar power | The Guardian
@thomasfuchs@hachyderm.ioRE: https://hachyderm.io/@thomasfuchs/116726575162112584
And even for the most optimal of use cases, code generation, which gives an LLM every possible advantage (huge amounts of self-similar training data with millions of iterations of the same problems and solutions, strictly deterministic symbol sequences in highly simplified languages), LLMs are starting to approach local maxima; they’re even regressing when context windows get too large (so more capable hardware doesn’t help).
I still strongly believe that they’re useless for 99% of tasks they’re marketed for and for the rest (like coding) there will be locally run solutions that are affordable.
¯\_(ツ)_/¯
@vosje62@mastodon.nlGoeie uitleg en context van Arjen vd Horst in 8uurjournaal over Belfast.
#8uurjournaal #Belfast
@aardrian@toot.cafeIn many cases, the people citing the LLM to me either won’t listen to why it’s wrong or don’t care.
Because their employee reviews are a function of how much they use the LLM versus accuracy, bug-free code, giving a shit, …
https://bsky.app/profile/scottohara.me/post/3mlwdps…
@jredlund@social.linux.pizzaThe Speakers
This is a spoken word poem read on top of a vaguely hip-hop backing track posted on the "Twanger Central" section of the guitar-oriented discussion board www.tdpri.com by a player whose screen name is "cminor9." The poem was a morning poem (sometimes I wake up with some lines) that I didn't know what to do with. It fit here. "Speakers" is a synonym for "humans" in this context.
@cdarwin@c.imDonald Trump needs a deal to save him in the midterms
more than he wants a nuclear deal
and whatever happens in the next few days should be understood in that context.
It’s why he is about to give Iran a ton of money to save face
as Trump’s approval ratings are as low as ever and our farmers are begging for fertilizer.
@arXiv_eessAS_bot@mastoxiv.pageEvaluating the Expressive Appropriateness of Speech in Rich Contexts
Tianrui Wang, Ziyang Ma, Yizhou Peng, Haoyu Wang, Zhikang Niu, Zikang Huang, Yihao Wu, Yi-Wen Chao, Yu Jiang, Yuheng Lu, Guanrou Yang, Xuanchen Li, Hexin Liu, Chunyu Qiang, Cheng Gong, Yifan Yang, Tianchi Liu, Junyu Wang, Nana Hou, Meng Ge, Fuming You, Wei Yang, Zhongqian Sun, Haifeng Hu, Xiaobao Wang, Eng Siong Chng, Xie Chen, Longbiao Wang, Jianwu Dang
https://arxiv.org/abs/2605.09413 https://arxiv.org/pdf/2605.09413 https://arxiv.org/html/2605.09413
arXiv:2605.09413v1 Announce Type: new
Abstract: Evaluating expressive speech remains challenging, as existing methods mainly assess emotional intensity and overlook whether a speech sample is expressively appropriate for its contextual setting. This limitation hinders reliable evaluation of speech systems used in narrative-driven and interactive applications, such as audiobooks and conversational agents. We introduce CEAEval, a Context-rich framework for Evaluating Expressive Appropriateness in speech, which assesses whether a speech sample expressively aligns with the underlying communicative intent implied by its discourse-level narrative context. To support this task, we construct CEAEval-D, the first context-rich speech dataset with real human performances in Mandarin conversational speech, providing narrative descriptions together with fifteen dimensions of human annotations covering expressive attributes and expressive appropriateness. We further develop CEAEval-M, a model that integrates knowledge distillation, planner-based multi-model collaboration, adaptive audio attention bias, and reinforcement learning to perform context-rich expressive appropriateness evaluation. Experiments on a human-annotated test set demonstrate that CEAEval-M substantially outperforms existing speech evaluation and analysis systems.
toXiv_bot_toot
@Techmeme@techhub.socialResearchers detail "context bombing", where defenders use prompt injections to trigger guardrails of attackers' LLMs, cutting AI hacking success rates by ~90% (Dan Goodin/Ars Technica)
https://arstechnica.com/security/2026/07/now-d…
@hippyjo@c.imUkraine update from Giorgio Provinciali: As a weapons development context, Ukraine offers unrivaled opportunity.
As an engineer, Giorgio distills the marketing speech into reality, and he shares the math.
Share the friend link to bypass the paywall, and help give the world a glimpse of the reality on the ground in Ukraine:
@memeorandum@universeodon.comDon't Forget the Broader Context of the Iranian Memorandum (Victor Davis Hanson/American Greatness)
https://amgreatness.com/2026/06/23/dont-forget-the-broader-context-of-the-iranian-memorandum/
http://www.memeorandum.com/260623/p86#a260623p86
@azonenberg@ioc.exchangeIf I say 1.3.6.1.4.1 with no context, do you instantly know what I'm talking about?
@Mediagazer@mstdn.socialAnalysis: the 2026 Emmy nominations for acting and hosting hit an 11-year low with only 22 performers of color recognized, down 21% from the nominees in 2025 (Anthony D'Alessandro/Deadline)
https://deadline.com/2026/07/2026-emmy-nominations-…
@frankel@mastodon.top @Techmeme@techhub.socialMoonshot AI releases Kimi K3, a 2.8T-parameter AI model that it says rivals Opus 4.8 and GPT 5.5, and plans to release model weights by July 27 (Kimi)
https://www.kimi.com/blog/kimi-k3
@zachleat@zachleat.comanyway, it’s particularly juicy that the original compromise came from trying to cheat at roblox? https://www.infostealers.com/article/breaking-vercel-breach-linked-to-infostealer-infection-at-context-ai/
@grahamperrin@bsd.cafe@… thanks, what's the #FreeBSD context?
@philip@mastodon.mallegolhansen.com@… I’ve greatly enjoyed Goldratt’s books on the Theory of Constraints in that context.
So much time is being wasted optimizing non-bottlenecks.
@burger_jaap@mastodon.socialThis is essential reading on competition in the European EV charging market, and on how Charging Point Operators (CPOs) cannot treat Mobility Service Providers (MSPs) and their own internal business units unequally. Regulators can already take effective action against market abuse even before #AFIR, the European Alternative Fuels Infrastructure Regulation is enforced more effectively
@shriramk@mastodon.socialThe closure of Hampshire College means the Five College consortium will have only four colleges…but do we have a name for that reduction? Yes, since 2008, when Stephen Chrisomalis (in another context) coined «quinquemation»!
https://glossographia.com/2008/11/30/five-pa…
@sofia@chaos.socialwithout context, how do you feel about the word "loyalty"?
@Techmeme@techhub.socialZhipu launches GLM-5.2, says it markedly improves coding and agentic tasks, with strong long-horizon capabilities and a 1M context window, under an MIT license (Z.ai)
https://z.ai/blog/glm-5.2
@lpryszcz@genomic.socialLet's try sth new: we've seen a beautiful house to buy in our neighborhood and we're considering putting an offer next week.
Are there any clear reasons we shouldn't do it now?
For the context, we don't have any debts and we are in rather stable employment (although you can never know for sure). We could get a #mortgage with fixed rate for 30 years.
@fanf@mendeddrum.orgfrom my link log —
Debootstrapping without archeology: stacked implementations in Camlboot.
https://arxiv.org/abs/2202.09231
saved 2026-06-10 https://d…
@berlinbuzzwords@floss.socialDavid Louis Hollembaek (Veeva Systems) and Hande Kafkas (cognee) will tackle one of the most pressing AI challenges right now, getting context right for AI agents. Two search veterans, one great session.
Learn more: https://2026.berlinbuzzwords.de/session/search-is-back-solving-the-context-crisis-for-ai-agents/
Get your ticket: https://2026.berlinbuzzwords.de/tickets/
@Techmeme@techhub.socialThreads is testing a Meta AI integration similar to X's Grok, letting users mention Meta AI in a post or a reply to get more context, in five countries (Aisha Malik/TechCrunch)
https://techcrunch.com/2026/05/12/threads-tests-a-meta-ai-i…
@gwire@mastodon.socialI recently made a passing reference to the Mars Climate Orbiter in the context of software testing, but I guess it's now far enough in the past that younger people don't know it.
(Short version: NASA specified that measurement values should be in SI (metric) units, their supplier delivered a system that used customary (imperial) units. NASA didn't adequately test against specification, and the probe got destroyed during orbital insertion.)
@michabbb@social.vivaldi.net🧩 Bonus: container attribute injection with #[Context('trace_id')], conditional when()/unless(), exception reporting that carries context, and survival across the scheduler subprocess via __LARAVEL_CONTEXT.
✅ Use it for trace IDs, tenants, route names, user IDs, locale. Not for required business inputs, secrets, or large payloads. If the code can't run without it, pass it explicitly.
🌐
@inthehands@hachyderm.ioAs a teaser, I’ll thread the 20 tech lessons from the talk here — without the context, but at least you’ll get the vibe.
–––
LESSON 1: Know what problem you are solving.
LESSON 2: Security requires threat modeling.
@cdarwin@c.imCamille Mumford / Emerson Polling:
California 2026 Poll: Becerra Continues to Surge, Steyer and Hilton Compete for Second Spot — Home Polls California 2026 Poll: Becerra Continues to Surge, Steyer and Hilton Compete for Second Spot — Bass leads primary for LA Mayor with 30% support; Pratt and Raman trail
https://www.
@grahamperrin@bsd.cafe @shriramk@mastodon.socialWeird that our simplest, textual interfaces are bringing the most amount of whimsy back into computing.
Good night, sleep Claude. Hope you catch some winks once you're done running my `/loop`. (Alt text for context.)
@cosmos4u@scicomm.xyz @netzschleuder@social.skewed.defacebook_friends: Maier Facebook friends (2014)
A small anonymized Facebook ego network, from April 2014. Nodes are Facebook profiles, and an edge exists if the two profiles are "friends" on Facebook. Metadata gives the social context for the relationship between ego and alter.
This network has 362 nodes and 1988 edges.
Tags: Social, Online, Unweighted, Metadata
@frankel@mastodon.topI continue to experiment with #AI in the context of #softwareengineering. I’m fortunate that my team supports me in exploring different ways to improve our daily work. This week, I designed a team of #autonomousagents
@mgorny@social.treehouse.systemsSure, "Gladiator" as a company name sounds cool at first, but have you considered the actual historical context of the term?
@Mediagazer@mstdn.socialMeta launches Muse Image in Meta AI, Instagram, and WhatsApp, and previews Muse Video, the first media generation models from its Superintelligence Labs (Meta)
https://ai.meta.com/blog/introducing-muse-image-muse-video-msl/
@sean@scoat.esRan into yet another case where doing things the new/currently blessed Apple Way fails because their new way doesn't actually have everything needed that the old way provided.
Tonight it's that I had to use XCTest instead of Swift Testing because I need the context to assert a macro expansion diagnostic. (…or maybe I’m just missing something…?)
But this applies elsewhere, too—especially with SwiftUI.
I know it can be hard, but I wish they'd provide complete solu…
@michabbb@social.vivaldi.net @cdarwin@c.imAnthropic has been trying to understand how large language models (LLMs) work for a few years now.
Anthropic isn’t the only one looking at this, but I think the company has made it part of its core mission more than most.
Anthropic’s CEO, Dario Amodei, has said we won’t be able to control LLMs fully unless we learn more about how they work.
So this new research is very much in that context.
It goes deeper into the weird mechanisms inside LLMs than ever before.
@fanf@mendeddrum.orgfrom my link log —
ARM LDM and STM are Turing complete.
https://kellanclark.github.io/2023/09/18/armfuck/
saved 2023-09-27 https://
@grahamperrin@bsd.cafe@… no, not only with SDDM.
<https://www.reddit.com/r/freebsd/comments/1t2jf6f/comment/ojsp2vy/?context=1>
"… it s…
@cdarwin@c.imFinancial Times:
ECB raises interest rates for first time since 2023
— The European Central Bank raised interest rates by a quarter point to 2.25 per cent on Thursday,
becoming the first central bank in the G7 to increase borrowing costs in response to Middle East energy crisis
https://www.memeor…
@michabbb@social.vivaldi.net🔍 Everything about Laravel's Context facade
Laravel's #Context facade is a framework-level place to store execution metadata during a request, command, or queued job — and attach it to your logs automatically. No more passing the same arrays through every method call.
🧠 Mental model: a per-execution key-value repository. Add data once (a trace_id), and any log later in the…
@Techmeme@techhub.socialGoogle DeepMind details a Gemini-powered mouse pointer that understands what it is pointing at, allowing users to perform tasks without using text-heavy prompts (Google DeepMind)
https://deepmind.google/blog/ai-pointer/
@Techmeme@techhub.socialOpenAI rolls out Chronicle, which builds memories from screen captures to make Codex more aware of context, as a research preview for Pro subscribers on macOS (Zac Hall/9to5Mac)
https://9to5mac.com/2026/04/20/codex-for-mac-gains…
@michabbb@social.vivaldi.net🧠 The reasoning: text tokens are discrete, drawn from a fixed vocabulary of ~50,000 entries, each mapping to a fixed point in embedding space.
Image tokens are continuous vectors of a thousand floating point numbers, so they can pack far more information into the same space.
🎮 The test: a tower-defense game where the AI plays in JSON mode without graphics, run for hours with two context variants — the same game state as a plain text string, and as an image rendered via PixelPipe.…
@frankel@mastodon.topI'm about to lose control, and I think I like it 🎶🎵
I'm so excited, and I just can't hide it 🎶🎵🎶🎶
https://blog.frankel.ch/java-geek-weekly/135/
@michabbb@social.vivaldi.net🕵️ Hidden context travels with execution but never appears in logs. Laravel uses it for unique queued jobs to store cache store and lock keys, releasing locks even when a model fails to unserialize.
🔒 Scoped context via Context::scope() adds metadata only for a block, restoring previous state afterward — even on exceptions. Prefer scalars and IDs; mutated objects can leak.
@Techmeme@techhub.socialHands-on with Siri AI: successfully executed multi-step prompts, understood context well, has strong guardrails, and is a bit more dispassionate than Gemini (Allison Johnson/The Verge)
https://www.theverge.com/tech/947432/siri-ai-apple-intelligence-ios-27-wwdc
@grahamperrin@bsd.cafe@… FreeBSD Installer for 15.1 will not include the desktop script, unfortunately.
<https://www.reddit.com/r/freebsd/comments/1t7m…
@grahamperrin@bsd.cafe… "not available or version mismatch" is puzzling for FreeBSD version 1501000. …
<https://www.reddit.com/r/freebsd/comments/1ugomc1/comment/ou7es29/?context=1>
Please, what am I missing?
@…
@Techmeme@techhub.socialOpenAI broadly releases GPT-5.6, and launches ChatGPT Work, an AI agent that can gather context across apps and files to create documents, on Mac and Windows (Axios)
https://www.axios.com/2026/07/09/ai-openai-gpt-release
@cdarwin@c.imJames Talarico ripped the brisket with his bare hands and polished off a rack of pork ribs
https://www.memeorandum.com/260605/p64#a260605p64
@michabbb@social.vivaldi.net🎯 Futures deliver deferred results for async computations, with composition via await_all and await_first. Context associates data with a coroutine for easy adaptation of global-variable code.
🗓️ Roadmap: Complete Async API (v0.6) shipped March 2026, with Threads & Stabilization in Summer 2026, Framework Adapters in Q3 2026, and a stable v1.0 release targeted for November 2026 on PHP 8.6.
🌐
@Techmeme@techhub.socialApple announces Siri AI, with on-screen awareness, personal context understanding, a Dynamic Island-based interface, and more, along with a new Siri app (Zac Hall/9to5Mac)
https://9to5mac.com/2026/06/08/new-siri-whats-new/
@Techmeme@techhub.socialApple announces new Apple Intelligence features for its apps, including a Notify Me feature in Safari and a Call Context feature in the Phone app (John Quintet/iPhone in Canada)
https://www.iphoneincanada.ca/2026/06/08/safari-ga…
@michabbb@social.vivaldi.net🔧 Three pieces mirror the Python RLM design: a system prompt, a $CONTEXT file plus bash as the REPL, and the native rlm_query sub-call. Depth 0 and 1 get full tools; the leaf at max depth runs bash only. jj workspaces isolate child edits when available.
🛡️ Five guardrails guarantee termination: dollar budget, wall-clock timeout, call limit, depth limit and PATH scrubbing. Deeper agents are told to be more conservative, preferring direct action over spawning more children.
@michabbb@social.vivaldi.net🚚 The real power: crossing the queue boundary. Context dehydrates into the job payload under illuminate:log:context, then hydrates back on JobProcessing — so an async job's logs carry the original trace_id and order_id.
🪝 Hooks dehydrating()/hydrated() restore values like locale across the queue. Modify the Repository passed to the callback, not the facade.
@Techmeme@techhub.socialAnthropic launches Claude Tag, an agentic AI coworker for Slack that can learn context, give suggestions, and more, in beta for Claude Team and Enterprise tiers (David Gewirtz/ZDNET)
https://www.zdnet.com/article/anthropic-claude-tag-agentic-ai-coworker-slack/…
@Techmeme@techhub.socialGoogle unveils Workspace Intelligence, which understands "complex semantic relationships" between data in Workspace apps to provide personalized context (Abner Li/9to5Google)
https://9to5google.com/2026/04/22/google-workspace-intelligence/
@Techmeme@techhub.socialAlibaba releases Qwen3.7-Plus, a multimodal proprietary model with a 1M-token context window, costing $2 per 1M tokens, 60% less than text-only Qwen3.7-Max (Carl Franzen/VentureBeat)
https://venturebeat.com/technology/a…
@Techmeme@techhub.socialxAI launches Grok 4.3, featuring "always-on reasoning", 1M token context window, and low API pricing, and releases a voice cloning suite called Custom Voices (Carl Franzen/VentureBeat)
https://venturebeat.com/technology/xai
