Tootfinder

Pinterest users, especially artists, says the platform has gotten worse in the past year due to AI moderation, AI-generated art, and AI features (Matthew Gault/404 Media)
https://www.404media.co/pinterest-is-drowning-in-a-sea-of-ai-slop-and-auto-moderat…

Pinterest Is Drowning in a Sea of AI Slop and Auto-Moderation
Users are exhausted fighting AI moderation, AI-generated art, and AI-first features.

This came in via #B3ta but hilariously it recalls one of my first generative compositions from well-before the computer era, my submission to some CBC call for composers and done the old-fashioned way, graph paper and pencil, I used Conway's Live the same way, but on tablature for classical guitar.
It was a bitch to play and I didn't win any award, but in late-70's CBC 'classical' sensibilities, I submitted it as a lark, although I was serious. Well, mostly: the second section of Conway's Life for Guitar used a starting pattern that would last the duration, but was then CONSTRAINED by a musical rules of harmony using Jerome Kern's All The Things You Are to assassinate non-conforming tones and indeed, as expected, I found the notion of Harmony to be highly toxic. No one likes a composer with a sense of humour 😆
So far as I know, no recordings of my entry exist, the score is likely somewhere in the bottom of my notorious trunk so hated every moving day.
Conway's Life Music
https://vovanz.github.io/conways-life-music/

Bikelution: Federated Gradient-Boosting for Scalable Shared Micro-Mobility Demand Forecasting
Antonios Tziorvas, Andreas Tritsarolis, Yannis Theodoridis
https://arxiv.org/abs/2602.20671 https://arxiv.org/pdf/2602.20671 https://arxiv.org/html/2602.20671
arXiv:2602.20671v1 Announce Type: new
Abstract: The rapid growth of dockless bike-sharing systems has generated massive spatio-temporal datasets useful for fleet allocation, congestion reduction, and sustainable mobility. Bike demand, however, depends on several external factors, making traditional time-series models insufficient. Centralized Machine Learning (CML) yields high-accuracy forecasts but raises privacy and bandwidth issues when data are distributed across edge devices. To overcome these limitations, we propose Bikelution, an efficient Federated Learning (FL) solution based on gradient-boosted trees that preserves privacy while delivering accurate mid-term demand forecasts up to six hours ahead. Experiments on three real-world BSS datasets show that Bikelution is comparable to its CML-based variant and outperforms the current state-of-the-art. The results highlight the feasibility of privacy-aware demand forecasting and outline the trade-offs between FL and CML approaches.
toXiv_bot_toot

«LLMs are cliché machines, trained on a resilient human weakness for generating maximum content with minimum effort.»
Bingo.
Unfortunately, this too hits the nail on the head: «Bad art is something human beings love to do, in vast numbers. It’s part of who we are, and when abandoned by inspiration we trust in the same methods we’ve programmed into LLMs.»

Sure, AI can ‘do’ writing. But memoir? Not so much | Aeon Essays
As AI’s endless clichés continue to encroach on human art, the true uniqueness of our creativity is becoming ever clearer

Very nice German/French documentary about the Alps as central Europe's weather & climate generator (available until end of May):
Die Alpen - Europas Wetterküche
https://www.arte.tv/de/videos/110976-000-A/die-alpen-europas-wetterkueche/

Adobe launches Firefly Custom Models in public beta, letting users train AI image generators on their own assets; the custom models are private by default (Jess Weatherbed/The Verge)
https://www.theverge.com/tech/897243/adobe-firefly-ai-custom-models-ima…

Adobe’s AI image generator can now be trained on your own art
Adobe is launching customizable AI image generators that can mimic specific artistic styles and character designs.

Adobe launches Firefly Custom Models in public beta, letting users train AI image generators on their own assets; the custom models are private by default (Jess Weatherbed/The Verge)
https://www.theverge.com/tech/897243/adobe-firefly-ai-custom-models-ima…

Adobe’s AI image generator can now be trained on your own art
Adobe is launching customizable AI image generators that can mimic specific artistic styles and character designs.

ProphetKV: User-Query-Driven Selective Recomputation for Efficient KV Cache Reuse in Retrieval-Augmented Generation
Shihao Wang, Jiahao Chen, Yanqi Pan, Hao Huang, Yichen Hao, Xiangyu Zou, Wen Xia, Wentao Zhang, Haitao Wang, Junhong Li, Chongyang Qiu, Pengfei Wang
https://arxiv.org/abs/2602.02579 https://arxiv.org/pdf/2602.02579 https://arxiv.org/html/2602.02579
arXiv:2602.02579v1 Announce Type: new
Abstract: The prefill stage of long-context Retrieval-Augmented Generation (RAG) is severely bottlenecked by computational overhead. To mitigate this, recent methods assemble pre-calculated KV caches of retrieved RAG documents (by a user query) and reprocess selected tokens to recover cross-attention between these pre-calculated KV caches. However, we identify a fundamental "crowding-out effect" in current token selection criteria: globally salient but user-query-irrelevant tokens saturate the limited recomputation budget, displacing the tokens truly essential for answering the user query and degrading inference accuracy.
We propose ProphetKV, a user-query-driven KV Cache reuse method for RAG scenarios. ProphetKV dynamically prioritizes tokens based on their semantic relevance to the user query and employs a dual-stage recomputation pipeline to fuse layer-wise attention metrics into a high-utility set. By ensuring the recomputation budget is dedicated to bridging the informational gap between retrieved context and the user query, ProphetKV achieves high-fidelity attention recovery with minimal overhead. Our extensive evaluation results show that ProphetKV retains 96%-101% of full-prefill accuracy with only a 20% recomputation ratio, while achieving accuracy improvements of 8.8%-24.9% on RULER and 18.6%-50.9% on LongBench over the state-of-the-art approaches (e.g., CacheBlend, EPIC, and KVShare).
toXiv_bot_toot

«LLMs are cliché machines, trained on a resilient human weakness for generating maximum content with minimum effort.»
Bingo.
Unfortunately, this too hits the nail on the head: «Bad art is something human beings love to do, in vast numbers. It’s part of who we are, and when abandoned by inspiration we trust in the same methods we’ve programmed into LLMs.»

US-based Credo, which specializes in data center connectivity, agrees to acquire Israeli chip company DustPhotonics in a cash-and-stock deal worth up to $1.3B (CTech)
https://www.calcalistech.com/ctechnews/article/d9pizceo2

Credo to acquire Israeli chip startup DustPhotonics for up to $1.3 billion
The Israeli company's photonic chip technology enables faster, lower-cost data transfer in next-generation AI clusters.