mastodon.world is one of the many independent Mastodon servers you can use to participate in the fediverse.
Generic Mastodon server for anyone to use.

Server stats:

8.4K
active users

#Diffusion

0 posts0 participants0 posts today
Unofficial PetaPixel Bot<p>Moment’s CineBloom VND Filter Combines Dreamy Diffusion and Exposure Control <a href="https://petapixel.com/2025/07/08/moments-cinebloom-vnd-filter-combines-dreamy-diffusion-and-exposure-control/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">petapixel.com/2025/07/08/momen</span><span class="invisible">ts-cinebloom-vnd-filter-combines-dreamy-diffusion-and-exposure-control/</span></a> <a href="https://toot.earth/tags/variableneutraldensity" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>variableneutraldensity</span></a> <a href="https://toot.earth/tags/cinematography" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>cinematography</span></a> <a href="https://toot.earth/tags/videography" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>videography</span></a> <a href="https://toot.earth/tags/Equipment" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Equipment</span></a> <a href="https://toot.earth/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a> <a href="https://toot.earth/tags/ndfilter" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ndfilter</span></a> <a href="https://toot.earth/tags/dreamy" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>dreamy</span></a> <a href="https://toot.earth/tags/filter" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>filter</span></a> <a href="https://toot.earth/tags/moment" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>moment</span></a> <a href="https://toot.earth/tags/News" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>News</span></a> <a href="https://toot.earth/tags/VND" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>VND</span></a> <a href="https://toot.earth/tags/nd" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>nd</span></a></p>
foldworks<p>This effect is new to me, done on a grey rectangle (about 50 iterations) in Affinity Photo. </p><p>Adding noise makes the effect quicker? And easier to do than simulating Turing-style diffusion, say.</p><p>h/t <a href="https://mastodon.gamedev.place/@rich/114810794227569284" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">mastodon.gamedev.place/@rich/1</span><span class="invisible">14810794227569284</span></a><br><a href="https://mathstodon.xyz/tags/Affinity" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Affinity</span></a> <a href="https://mathstodon.xyz/tags/AffinityPhoto" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AffinityPhoto</span></a> <a href="https://mathstodon.xyz/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a> <a href="https://mathstodon.xyz/tags/TextureTuesday" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>TextureTuesday</span></a></p>
Hacker News<p>Mercury: Ultra-Fast Language Models Based on Diffusion</p><p><a href="https://arxiv.org/abs/2506.17298" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">arxiv.org/abs/2506.17298</span><span class="invisible"></span></a></p><p><a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HackerNews</span></a> <a href="https://mastodon.social/tags/Mercury" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Mercury</span></a> <a href="https://mastodon.social/tags/Ultra" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Ultra</span></a>-Fast <a href="https://mastodon.social/tags/Language" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Language</span></a> <a href="https://mastodon.social/tags/Models" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Models</span></a> <a href="https://mastodon.social/tags/Diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Diffusion</span></a> <a href="https://mastodon.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://mastodon.social/tags/Research" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Research</span></a></p>
Agenda du Libre<p>Montpellier: Émission | Radio FM-Plus | Temps Libre | Diffusion, Le lundi 7 juillet 2025 de 08h00 à 09h00. <a href="https://www.agendadulibre.org/events/32744" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="">agendadulibre.org/events/32744</span><span class="invisible"></span></a> <a href="https://pouet.chapril.org/tags/montpelLibre" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>montpelLibre</span></a> <a href="https://pouet.chapril.org/tags/radio" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>radio</span></a> <a href="https://pouet.chapril.org/tags/fmPlus" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>fmPlus</span></a> <a href="https://pouet.chapril.org/tags/tempsLibre" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>tempsLibre</span></a> <a href="https://pouet.chapril.org/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a></p>
Habr<p>Часть 3: Diffusion Transformer (DiT) — Stable Diffusion 3 как она есть</p><p>В этой статье погрузимся в мир генерации изображений с Diffusion Transformer (DiT) — сердцем Stable Diffusion 3. Разберем как она устроена и как работает</p><p><a href="https://habr.com/ru/articles/924410/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">habr.com/ru/articles/924410/</span><span class="invisible"></span></a></p><p><a href="https://zhub.link/tags/stable_diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>stable_diffusion</span></a> <a href="https://zhub.link/tags/transformer" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>transformer</span></a> <a href="https://zhub.link/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a> <a href="https://zhub.link/tags/vae" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>vae</span></a></p>
Marcel Göertz<p>Another recent creation. I love how diffusion has become a tool to visualize ideas, to quickly transform the noise in my head into a quick draft, a sketch. This is one of the ideas that was stuck in my head. <a href="https://mas.to/tags/ballet" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ballet</span></a> <a href="https://mas.to/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a></p>
Agenda du Libre<p>Montpellier: Émission | Radio FM-Plus | Temps Libre | Diffusion, Le lundi 23 juin 2025 de 09h00 à 10h00. <a href="https://www.agendadulibre.org/events/32702" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="">agendadulibre.org/events/32702</span><span class="invisible"></span></a> <a href="https://pouet.chapril.org/tags/montpelLibre" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>montpelLibre</span></a> <a href="https://pouet.chapril.org/tags/radio" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>radio</span></a> <a href="https://pouet.chapril.org/tags/fmPlus" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>fmPlus</span></a> <a href="https://pouet.chapril.org/tags/tempsLibre" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>tempsLibre</span></a> <a href="https://pouet.chapril.org/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a></p>
Claude Trudel<p>Netflix commencera à diffuser des chaînes de diffusion traditionnelles françaises l'été prochain<br><a href="https://arstechnica.com/gadgets/2025/06/netflix-will-start-showing-traditional-broadcast-channels-next-summer/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">arstechnica.com/gadgets/2025/0</span><span class="invisible">6/netflix-will-start-showing-traditional-broadcast-channels-next-summer/</span></a><br><a href="https://jasette.facil.services/tags/TV" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>TV</span></a> <a href="https://jasette.facil.services/tags/t%C3%A9l%C3%A9vision" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>télévision</span></a> <a href="https://jasette.facil.services/tags/TF1" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>TF1</span></a> <a href="https://jasette.facil.services/tags/USA" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>USA</span></a> <a href="https://jasette.facil.services/tags/France" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>France</span></a> <a href="https://jasette.facil.services/tags/Netfkix" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Netfkix</span></a> <a href="https://jasette.facil.services/tags/%C3%A9missions" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>émissions</span></a> <a href="https://jasette.facil.services/tags/continu" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>continu</span></a> <a href="https://jasette.facil.services/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a></p>
deepseekWhy would anyone let LLMs predict 4 tokens at once? The field of natural language processing (NLP) is witnessing another paradigm shift which could make its current methods seem quaint in the years...<br><br><a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/AI" target="_blank">#AI</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Large" target="_blank">#Large</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Language" target="_blank">#Language</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Models" target="_blank">#Models</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Natural" target="_blank">#Natural</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Language" target="_blank">#Language</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Processing" target="_blank">#Processing</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/DeepSeek" target="_blank">#DeepSeek</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/V3" target="_blank">#V3</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Diffusion" target="_blank">#Diffusion</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Language" target="_blank">#Language</a><br><br><a href="https://www.franksworld.com/2025/06/19/why-would-anyone-let-llms-predict-4-tokens-at-once/?utm_source=rss&amp;utm_medium=rss&amp;utm_campaign=why-would-anyone-let-llms-predict-4-tokens-at-once" rel="nofollow noopener" target="_blank">Origin</a> | <a href="https://awakari.com/sub-details.html?id=deepseek" rel="nofollow noopener" target="_blank">Interest</a> | <a href="https://awakari.com/pub-msg.html?id=K4jpsX7ymM0I1jbiK1oxGlVqBeq&amp;interestId=deepseek" rel="nofollow noopener" target="_blank">Match</a>
Hacker News<p>Rethinking Losses for Diffusion Bridge Samplers</p><p><a href="https://arxiv.org/abs/2506.10982" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="">arxiv.org/abs/2506.10982</span><span class="invisible"></span></a></p><p><a href="https://mastodon.social/tags/HackerNews" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>HackerNews</span></a> <a href="https://mastodon.social/tags/Rethinking" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Rethinking</span></a> <a href="https://mastodon.social/tags/Losses" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Losses</span></a> <a href="https://mastodon.social/tags/Diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Diffusion</span></a> <a href="https://mastodon.social/tags/Bridge" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Bridge</span></a> <a href="https://mastodon.social/tags/Samplers" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Samplers</span></a> <a href="https://mastodon.social/tags/MachineLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>MachineLearning</span></a> <a href="https://mastodon.social/tags/Research" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Research</span></a> <a href="https://mastodon.social/tags/Arxiv" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>Arxiv</span></a></p>
LLMs<br><br><a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/AI" target="_blank">#AI</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/AI," target="_blank">#AI,</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/ML" target="_blank">#ML</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/and" target="_blank">#and</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Deep" target="_blank">#Deep</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/Learning" target="_blank">#Learning</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/autoregression" target="_blank">#autoregression</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/DeepMind" target="_blank">#DeepMind</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/diffusion" target="_blank">#diffusion</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/diffusion" target="_blank">#diffusion</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/language" target="_blank">#language</a><br><br><a href="https://venturebeat.com/ai/beyond-gpt-architecture-why-googles-diffusion-approach-could-reshape-llm-deployment/" rel="nofollow noopener" target="_blank">Origin</a> | <a href="https://awakari.com/sub-details.html?id=LLMs" rel="nofollow noopener" target="_blank">Interest</a> | <a href="https://awakari.com/pub-msg.html?id=F5LhVIDiVeT5H0SzOMeBSN1m2qm&amp;interestId=LLMs" rel="nofollow noopener" target="_blank">Match</a>
Nicole Sharp<p><strong>Proving Superdiffusion</strong></p><p>Turbulence is very good at spreading things out. Drop dye into a turbulent flow and it will quickly disperse. Add in particles — like rubber ducks — and they can spread apart, often at speeds quicker than one would expect, based on the background flow. This is (roughly speaking) a phenomenon known as “<a href="https://en.wikipedia.org/wiki/Anomalous_diffusion" rel="nofollow noopener" target="_blank">superdiffusion</a>,” where turbulence makes particles that start out as neighbors part ways.</p><p>Physicists conjectured that turbulence — including simplified and idealized versions of it that are simpler to deal with — had this superdiffusion property, but no one was able to show that in a mathematically rigorous way. But now a group of mathematicians has done so, using a technique known as homogenization. There’s a lot more on the story over at <a href="https://www.quantamagazine.org/new-superdiffusion-proof-probes-the-mysterious-math-of-turbulence-20250516/?__readwiseLocation=" rel="nofollow noopener" target="_blank">Quanta</a>, or you can check out the original papers on arXiv. (Image credit: <a href="https://unsplash.com/photos/yellow-and-red-plastic-toy-VTvnoNBowZs" rel="nofollow noopener" target="_blank">J. Richard</a>; research credit: <a href="https://arxiv.org/abs/2404.01115" rel="nofollow noopener" target="_blank">S. Armstrong et al.</a> and <a href="https://arxiv.org/abs/2405.10732" rel="nofollow noopener" target="_blank">S. Armstrong and T. Kuusi</a>; see also <a href="https://www.quantamagazine.org/new-superdiffusion-proof-probes-the-mysterious-math-of-turbulence-20250516/?__readwiseLocation=" rel="nofollow noopener" target="_blank">Quanta</a>)</p><p><a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://fyfluiddynamics.com/tagged/diffusion/" target="_blank">#diffusion</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://fyfluiddynamics.com/tagged/fluid-dynamics/" target="_blank">#fluidDynamics</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://fyfluiddynamics.com/tagged/mathematics/" target="_blank">#mathematics</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://fyfluiddynamics.com/tagged/physics/" target="_blank">#physics</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://fyfluiddynamics.com/tagged/science/" target="_blank">#science</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://fyfluiddynamics.com/tagged/turbulence/" target="_blank">#turbulence</a></p>
2ra1Tf9fLmRR5RHXI4JbzLRiU6xНевероятный искусственный интеллект Easy Diffusion&nbsp;3.0 "Р...<br><br><br><a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/stable" target="_blank">#stable</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/stable" target="_blank">#stable</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/diffusion" target="_blank">#diffusion</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/stablediffusion" target="_blank">#stablediffusion</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/stable" target="_blank">#stable</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/diffusion" target="_blank">#diffusion</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/нейросеть" target="_blank">#нейросеть</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/stable" target="_blank">#stable</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/diffuison" target="_blank">#diffuison</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/stable" target="_blank">#stable</a> <a rel="nofollow noopener" class="mention hashtag" href="https://mastodon.social/tags/diffusion" target="_blank">#diffusion</a><br><a href="https://habr.com/ru/articles/916980/?utm_source=habrahabr&amp;utm_medium=rss&amp;utm_campaign=916980" rel="nofollow noopener" target="_blank">Origin</a> | <a href="https://awakari.com/sub-details.html?id=2ra1Tf9fLmRR5RHXI4JbzLRiU6x" rel="nofollow noopener" target="_blank">Interest</a> | <a href="https://awakari.com/pub-msg.html?id=CQeiuZQd6qqs3lJt3BtI01RXaWe&amp;interestId=2ra1Tf9fLmRR5RHXI4JbzLRiU6x" rel="nofollow noopener" target="_blank">Match</a>
Mark Carrigan<p><strong>The doom loops of generative&nbsp;AI</strong></p><pre><em><strong>This is a Claude Opus 4 summary of my recent slide decks and draft writing, in order to help me better understand what I'm trying to say with the notion of 'doom loops'</strong></em></pre><p>The concept of a ‘doom loop’ captures a particularly pernicious form of collective action problem emerging from digital transformation. While individual actors make rational decisions to adopt new technologies to solve immediate problems, these decisions aggregate into systemic changes that worsen the very conditions they were meant to address. This dynamic is becoming increasingly visible in professional contexts where generative AI is being rapidly adopted.</p><p><strong>Defining the Doom Loop</strong></p><p>A doom loop operates through four interconnected mechanisms:</p><p><strong>1. Individual Rationality/Collective Irrationality</strong> Faced with competitive pressures, individuals adopt technological solutions that provide immediate advantages. An academic using ChatGPT to increase publication output acts rationally given tenure requirements and job market pressures. However, when this behavior scales across the profession, it ratchets up productivity expectations for everyone, creating an arms race that leaves all participants worse off.</p><p><strong>2. Temporal Displacement</strong> The benefits of adoption are immediate and tangible (reduced workload, increased output), while the costs are delayed and diffuse (degraded professional standards, automated replacement). This temporal structure makes it nearly impossible to resist adoption, even when actors understand the long-term consequences.</p><p><strong>3. Infrastructure Capture</strong> The platforms and tools that might enable collective resistance become part of the acceleration. Social media platforms that could facilitate professional organization are simultaneously the training data for AI systems. The infrastructure of communication becomes the infrastructure of replacement.</p><p><strong>4. Legitimacy Erosion</strong> As automated systems take over core professional functions, they undermine the basis for professional authority. When AI can produce academic papers or make diagnostic decisions, it becomes harder to justify why human judgment remains necessary. The profession’s adoption of these tools paradoxically validates their eventual replacement.</p><p><strong>The Unbundling Dynamic</strong></p><p>The doom loop operates through what we might call ‘functional unbundling’. Rather than wholesale replacement of workers, roles are decomposed into discrete functions, with the ‘routine’ elements automated while humans manage the systems. This appears to preserve employment while fundamentally transforming its nature.</p><p>An academic’s role traditionally bundled together research, teaching, mentoring, and service. AI enables these to be separated: automated grading systems, chatbot advisors, AI-generated lecture content. The academic becomes a quality controller rather than an educator. This isn’t efficiency—it’s the systematic stripping away of what Pasquale calls “specifically human powers.”</p><p><strong>Acceleration Through Crisis</strong></p><p>Financial crises accelerate doom loops by making short-term solutions irresistible. UK universities facing budget constraints see AI as a path to maintaining operations with fewer staff. The 92 institutions currently implementing redundancy programs create perfect conditions for this dynamic: urgent financial pressure meets technological solutionism.</p><p>This creates a ratchet effect. Once some institutions adopt AI to cut costs, competitive pressures force others to follow. The baseline shifts, and what was once unthinkable (automated essay grading, AI teaching assistants) becomes standard practice. Each crisis becomes an opportunity to further entrench automated systems.</p><p><strong>The Collective Action Paradox</strong></p><p>Traditional collective action problems assume actors could coordinate if transaction costs were low enough. The doom loop presents a crueler paradox: the tools that lower coordination costs are themselves part of the problem. Professional communities might organize on LinkedIn or Twitter, but these platforms are training data for the systems that will replace them.</p><p>Moreover, the secrecy surrounding AI adoption—Mollick’s “secret cyborgs”—prevents even basic coordination. Without transparency about who is using what tools and how, professional communities cannot develop coherent responses. The shame and uncertainty around AI use atomizes potential resistance.</p><p><strong>Breaking the Loop: Theoretical Requirements</strong></p><p>Escaping a doom loop requires more than individual resistance or better policies. It demands:</p><p><strong>1. Temporal Reframing</strong>: Making long-term costs visible and immediate. This might involve professional bodies creating metrics that capture human value rather than just productivity.</p><p><strong>2. Collective Standards</strong>: Moving beyond individual ethics to professional norms that shape the boundary between human and machine decision-making. This isn’t luddism but careful delineation of where human judgment remains irreducible.</p><p><strong>3. Infrastructure Alternatives</strong>: Building communication and organization systems that aren’t simultaneously feeding the replacement machinery. This might mean returning to older forms of professional organization or creating new platforms with different logics.</p><p><strong>4. Value Articulation</strong>: Explicitly theorizing and defending what makes human professional judgment valuable beyond its functional outputs. This means moving past efficiency arguments to questions of meaning, responsibility, and social value.</p><p><strong>Conclusion</strong></p><p>The doom loop framework reveals how technological transformation can create self-reinforcing cycles of degradation even when each individual decision appears rational. Understanding these dynamics is essential for professions grappling with AI adoption. The choice isn’t between embracing or rejecting these technologies, but understanding how our collective responses shape the futures we create.</p><p>The millions of small decisions Pasquale identifies aren’t just about individual tool use—they’re about whether we allow efficiency logics to determine professional futures or insist on preserving space for human judgment, creativity, and meaning. The doom loop isn’t inevitable, but breaking it requires seeing beyond our individual circumstances to the collective dynamics we’re creating.</p><p></p><p><a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://markcarrigan.net/tag/diffusion/" target="_blank">#diffusion</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://markcarrigan.net/tag/doom-loops/" target="_blank">#doomLoops</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://markcarrigan.net/tag/llms/" target="_blank">#LLMs</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://markcarrigan.net/tag/organisation/" target="_blank">#organisation</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://markcarrigan.net/tag/organisational-sociology/" target="_blank">#organisationalSociology</a> <a rel="nofollow noopener" class="hashtag u-tag u-category" href="https://markcarrigan.net/tag/technology/" target="_blank">#technology</a></p>
Agenda du Libre<p>Montpellier: Émission | Radio FM-Plus | Temps Libre | Diffusion, Le lundi 9 juin 2025 de 09h00 à 10h00. <a href="https://www.agendadulibre.org/events/32674" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="">agendadulibre.org/events/32674</span><span class="invisible"></span></a> <a href="https://pouet.chapril.org/tags/montpelLibre" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>montpelLibre</span></a> <a href="https://pouet.chapril.org/tags/radio" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>radio</span></a> <a href="https://pouet.chapril.org/tags/fmPlus" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>fmPlus</span></a> <a href="https://pouet.chapril.org/tags/tempsLibre" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>tempsLibre</span></a> <a href="https://pouet.chapril.org/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a></p>
Habr<p>Kandinsky 4.1 Image – новый генератор изображений от Сбера</p><p>В декабре прошлого года на конференции AI Journey наша команда представила диффузионную генеративную модель Kandinsky 4.0 Video. Теперь мы рады представить новую версию нашей модели генерации изображений Kandinsky 4.1 Image!</p><p><a href="https://habr.com/ru/companies/sberbank/articles/915760/" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">habr.com/ru/companies/sberbank</span><span class="invisible">/articles/915760/</span></a></p><p><a href="https://zhub.link/tags/%D0%B3%D0%B5%D0%BD%D0%B5%D1%80%D0%B0%D1%86%D0%B8%D1%8F_%D0%B8%D0%B7%D0%BE%D0%B1%D1%80%D0%B0%D0%B6%D0%B5%D0%BD%D0%B8%D0%B9" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>генерация_изображений</span></a> <a href="https://zhub.link/tags/kandinsky_4" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>kandinsky_4</span></a> <a href="https://zhub.link/tags/sberai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>sberai</span></a> <a href="https://zhub.link/tags/generative_models" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>generative_models</span></a> <a href="https://zhub.link/tags/texttoimage" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>texttoimage</span></a> <a href="https://zhub.link/tags/computer_vision" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>computer_vision</span></a> <a href="https://zhub.link/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a> <a href="https://zhub.link/tags/sft" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>sft</span></a> <a href="https://zhub.link/tags/artificial_intelligence" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>artificial_intelligence</span></a> <a href="https://zhub.link/tags/machine_learning" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>machine_learning</span></a></p>
Harald Klinke<p>🚀 {diffuseR} brings diffusion models to R — no Python required.<br>Generate images from text or modify existing ones using Stable Diffusion 2.1 and SDXL.</p><p>🖼️ 100% R-native<br>💻 Works on CPU and GPU<br>🌱 Contributions welcome!</p><p>🔗 <a href="https://github.com/cornball-ai/diffuseR" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://</span><span class="ellipsis">github.com/cornball-ai/diffuse</span><span class="invisible">R</span></a><br><a href="https://det.social/tags/rstats" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>rstats</span></a> <a href="https://det.social/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a> <a href="https://det.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>AI</span></a> <a href="https://det.social/tags/StableDiffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>StableDiffusion</span></a> <a href="https://det.social/tags/torch" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>torch</span></a> <a href="https://det.social/tags/OpenSource" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>OpenSource</span></a> <a href="https://det.social/tags/ImageGeneration" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ImageGeneration</span></a> <a href="https://det.social/tags/GenerativeAI" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>GenerativeAI</span></a></p>
Agenda du Libre<p>Montpellier: Émission | Radio FM-Plus | Temps Libre | Diffusion, Le lundi 2 juin 2025 de 09h00 à 10h00. <a href="https://www.agendadulibre.org/events/32643" rel="nofollow noopener" translate="no" target="_blank"><span class="invisible">https://www.</span><span class="">agendadulibre.org/events/32643</span><span class="invisible"></span></a> <a href="https://pouet.chapril.org/tags/montpelLibre" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>montpelLibre</span></a> <a href="https://pouet.chapril.org/tags/radio" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>radio</span></a> <a href="https://pouet.chapril.org/tags/fmPlus" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>fmPlus</span></a> <a href="https://pouet.chapril.org/tags/tempsLibre" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>tempsLibre</span></a> <a href="https://pouet.chapril.org/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a></p>
michabbb<p>okay, there is more speed when asking the right questions 😉 <a href="https://social.vivaldi.net/tags/google" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>google</span></a> <a href="https://social.vivaldi.net/tags/gemini" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>gemini</span></a> <a href="https://social.vivaldi.net/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a> <a href="https://social.vivaldi.net/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ai</span></a> <a href="https://social.vivaldi.net/tags/llm" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>llm</span></a></p>
michabbb<p>My first impression of <a href="https://social.vivaldi.net/tags/google" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>google</span></a> <a href="https://social.vivaldi.net/tags/gemini" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>gemini</span></a> <a href="https://social.vivaldi.net/tags/diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>diffusion</span></a> </p><p>110 tokens/s is "okay".... but using any <a href="https://social.vivaldi.net/tags/LLM" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>LLM</span></a> with such an old knowledge cutoff - is - for most cases - useless... as least for me..... compared to <a href="https://social.vivaldi.net/tags/groq" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>groq</span></a> with a cutoff of July 2024 and twice the speed... 🤔 it´s an experiment, I know.....</p><p><a href="https://social.vivaldi.net/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#<span>ai</span></a></p>