iCreatAI News
Back to main site
Skip to content
#Data / training
Nov 5Thu
  1. The Verge · AI58

    Big Tech’s big AI buildout will not end until someone stops it

    Amazon, Microsoft, Alphabet and Meta told investors they plan to spend at least $720 billion on capital expenditures this year, mostly on AI infrastructure, on top of at least $1 trillion in commitments for coming years, even though they have not yet made money from AI. Alphabet posted its worst one-day stock drop, sinking 13%, after acknowledging that Google Cloud's AI-driven growth came at the expense of its core Search business, where ad revenue growth slowed to 12% from 17% a year earlier. Microsoft was the only one of the four to avoid Wall Street's ire, reporting $45.2 billion in quarterly capex, below the expected $50 billion, due to third-party capacity delays, and it later announced a $25 billion bond sale.

  2. Hacker News AI22

    Exploring AI Contextual Coherence Across 500K+ Words and 1,788 Messages

    This deposit is a pseudonymised export of a single unplanned six-day conversation (19–25 October 2026) between a user and DeepSeek's chat model, which began as an analysis of a personal essay on forgiveness and grew into an extended exchange on psychological meta-analysis, a cumulative list of 143 analytical labels the model called "domains," family-history research, and a comparison of two short texts from 2012 and 2026. The conversation tree holds 1,788 message nodes, with the final branch containing 742 user turns and 746 model turns, about 559,000 words, 52 model-initiated web searches, and 15 image uploads, plus 296 nodes on edited or regenerated branches preserved in the JSON. The archive has been redacted to remove personal and identifying information, and the author frames it as documenting human-directed inquiry and AI-generated conceptual synthesis while noting limitations of AI-generated analysis regarding factual accuracy, scientific validation, and evidentiary reliability.

  3. MarkTechPost39

    Google AI Research Releases Agile Deliberation Framework for Gray-Area Image Classifiers

    Google Research released Agile Deliberation, a human-in-the-loop framework that helps users turn vague concepts like 'healthy food' into structured written definitions that also serve as prompts for a Gemini 2.5 Flash VLM classifier. It finds borderline images by meaning rather than classifier confidence, using borderline queries, CLIP/FAISS retrieval, dictionary-learning clusters and a UCB bandit, and averaged 10.5 F1 points above zero-shot Gemini 2.5 Flash across 2 concepts and 18 live sessions. The code is Apache 2.0 on GitHub with a demo notebook, and the work appears in a CVPR 2026 paper.

  4. The Verge · AI42

    Qualcomm is betting big on data centers as smartphone sales sink

    With smartphone sales down about 20 percent versus 2025, Qualcomm said on its Q4 2026 earnings call that it expects data center chips to eventually account for around 30 percent of its business, aided by partnerships with Amazon and Meta announced earlier this year. The company forecasts roughly $650 million in data center revenue in Q1 2027, rising to $5 billion by the end of 2027 and $15 billion by 2029, against total 2026 revenue of $10.4 billion.

  7. Tomasz Tunguz58

    Watts in Store for 2027 : The AI Ceiling

    Tomasz Tunguz argues the 2027 AI compute ceiling is about 80GW, roughly doubling all compute built in US history in twelve months, yet GPU hourly rental rates have already jumped from $3 to $12-13 and Anthropic's blended GPU-hour cost rose 82% in two quarters. He cites Brad Gerstner estimating 43GW coming online in 2027 against 38GW cumulative at end-2026, with 10-15GW of expected build possibly not coming online, and projects today's 38GW fleet earning about $140b while the 2027 additions deliver roughly $350b for nearly $500b total. Tunguz concludes that rising GPU prices may push foundation labs toward inference over training, make latency a release valve, encourage algorithmic innovation and harness efficiency, and force a reality check on vibe-coding economics.

  8. Databricks Blog42

    The Enterprise Case for Open-Weight Models on Databricks

    Databricks field engineers describe seven pressures pushing enterprises away from single frontier-model providers, including insufficient returns on AI spend, frequent ranking changes, guardrail refusals, shifting data-retention policies, governance demands, harness effects on performance, and open-weight models now benchmarking comparably to frontier models at lower cost. They present Unity Gateway as a unified API for routing and governing open-weight, frontier, and external models, citing internal data showing average unique models per customer rose from 4.2 to 7.1 since January, with roughly 40% open-weight. The post also notes GLM 5.2 reaching about 40% of open-weight tokens on Unity Gateway a month after launch and Databricks' $3 billion July raise, partly for GPU capacity to host open models.

  9. Latent Space62

    [AINews] Behind Latent Space: Kill My SaaS, Jev & AI in 2026 | AIEi Shanghai

    This AI News roundup covers the dispute over the "Weight Space Programming" story, noting that some claims were presented too favorably but that the results have been independently replicated, alongside an AI Engineer Shanghai feature and an interview on the history of Latent Space. It also recaps model releases and evals, including Tencent Hy4, Sarvam Prism, Kimi K3.1 Pro and LongCat-2.5, plus Anthropic's open-sourced FlashProve provable-inference system and benchmark audits from Prime Intellect and Xeophon.

  10. MIT Technology Review · AI67

    Texas was the new epicenter of the AI data center buildout. That might be kaput.

    Democrats won Texas' US Senate and governor races and took control of the state House for the first time in 24 years, with data centers a central campaign issue; governor-elect Gina Hinojosa pushed for a full moratorium after previously supporting tax breaks for data centers in 2023. The article notes massive projects underway, including the 1,820-megawatt Stargate Frontier in Shackelford County and the 6,400-megawatt Nexus Hubbard intended mostly to serve Anthropic, and that Governor Abbott directed a key state regulator on September 21 to halt data center permits. It adds that no single state office decides project outcomes, so local officials and construction unions may complicate follow-through on a moratorium.

  11. Hugging Face Blog60

    Dharma Field Extractor: Any Document In, Structured Fields in JSON Out

    Dharma-AI introduces the Dharma Field Extractor, a 2-billion-parameter model built on DharmaOCR-LITE v1.1 with a LoRA adapter trained on 43,384 examples, which takes a document image plus either an open request or an exact schema and returns typed JSON, including null for fields not present. On a 546-document, 1,638-case benchmark it scored 0.834 field-level F1, above every frontier configuration tested (Claude Opus 5 at 0.816 to GPT-5.6 Luna at 0.741), while running self-hosted on a single L40S at a cost the company reports as 6 to 147 times lower than the APIs compared. The company frames the result as evidence that a small model trained close to a bounded task can beat a general frontier model on that task.

    Why it matters: Useful for teams weighing self-hosted specialized extraction against frontier APIs, though the benchmark and cost figures come from the vendor itself.

  12. Google Research52

    Agile Deliberation: How AI helps humans define the gray areas

    Google Research introduces Agile Deliberation, a framework that treats building subjective AI classifiers as an iterative conversation, splitting the task into concept scoping (breaking a broad idea into sub-concepts and exclusions) and concept iteration (using vision-language models to surface borderline images that challenge the user's definitions). In 18 live 1.5-hour user sessions across the subjective domains of "Healthy Food" and "Pay-to-Play" content moderation, the approach produced more consistent, higher-performing classifiers than zero-shot, automated collaborator, and manual deliberation baselines while reducing user fatigue and anxiety. The code and algorithms are open-sourced on GitHub, with a Colab/Jupyter demo available.

  13. AWS Machine Learning Blog38

    How Indegene is modernizing and scaling enterprise data quality in life sciences

    Indegene built an LLM-powered Data Quality Agent within its NEXT Unified Data Platform on AWS, using Amazon Bedrock, Strands Agents, and Amazon Bedrock AgentCore to generate context-aware data quality rules, validate them, and produce SQL and Python code under human-in-the-loop governance. The agent draws on internal enterprise and external reference knowledge bases via RAG to recommend technical and life sciences-specific business rules. Indegene reports the agent reduced rule definition and implementation effort from about 30 hours to about 4 hours per dataset and cut new table onboarding from 2–5 working days to 2–4 hours.

  14. Nathan Lambert Interconnects38

    How big is the open model economy today?

    The author presents research notes from the Interconnects Dashboard, an aggregator of open model adoption and performance data, arguing that existing measurements of open model adoption are highly imprecise due to noisy platform metrics, startup marketing figures, and imperfect public evaluations. Using open model inference startups Baseten, Fireworks, and Together AI as bellwethers, along with OpenRouter's reported open model token share, the notes show near-exponential growth in disclosed token volume, with Fireworks reporting more than 40T tokens per day served to customers' customized models. The author overlays these figures against Google's reported platform and API token volumes from earnings reports to gauge how open model inference compares with closed and hyperscale providers.

  15. Hacker News AI48

    Show HN: First Silicon, real chip bugs that AI models miss

    First Silicon is a benchmark of 86 engineer fixes in open-source chip designs (84 real bugs, 70 from OpenTitan) in which models write SystemVerilog assertions judged by Yosys and SymbiYosys on both buggy and fixed versions. The best configuration, Claude Opus 5.5 at maximum effort, caught 16 of 86 cases with zero false alarms, and 67 of the 84 genuine fixes were caught cleanly by none of the four models at default settings. Letting Opus 5.5 call the bounded model checker did not close the gap, as the model dropped or changed 79% of counterexampled assertions, including 8 that had correctly exposed a bug.

  16. The Decoder85

    GPT-6 Astra sees the matrix and reprograms other AI models directly through their weights

    Researchers describe "Weight-Space Programming," in which a language model called Astra reads and edits the raw weights of other networks without step-by-step training: it identified which of four games a network was trained to play, generated untrained game-playing weights that beat a standard baseline in Breakout (517 vs. 25 points), and changed 0.016 percent of Qwen3.5-0.8B's weights to raise MATH-500 accuracy from 7.9 to 48.1 percent. It could not measurably improve Gemma 3 270M, and Jürgen Schmidhuber called the approach "old hat," while researcher Omar Khattab said the novelty lies in how well today's language models translate between human language and a network's numerical values. Experiment logs are on GitHub and the edited Qwen model is on Hugging Face.

    Why it matters: It offers a concrete look at weight-level model editing, including benchmark numbers and a public dispute over how novel the idea really is.

  17. Allen Institute for AI Blog38

    Ai2 and Google collaborate on a new generation of climate models

    Ai2 is launching a climate modeling initiative with multimillion-dollar, three-year funding from Google.org and technical support from Google Research to develop NeuralGCM+, an open-source model extending Google Research's NeuralGCM to simulate atmosphere and ocean together for faster, more reliable decade-scale climate projections. Ai2 will focus on climate trends over years to decades, such as how rainfall and water availability could change through mid-century in regions like the southwestern United States, while Google Research continues developing NeuralGCM for nearer-term forecasts over weeks to months. The work will proceed along three fronts: connecting the atmosphere and ocean, improving the response to a warming world, and making the tools easier to use and extend, including improved performance on widely used GPUs.

Nov 4Wed
  1. Hacker News AI36

    How to make scientific data archives AI-ready

    The article argues that scientific data archives are poorly suited to AI use because data is locked in opaque files, and proposes a four-step architecture: place data in scalable object storage, index chunks rather than filenames, co-locate dataset-level metadata with the data in object storage, and catalog repositories rather than individual files. It says this approach lets agents and humans quickly find and fetch only the chunks a query needs, citing a 576x reduction in data fetched for a GOES-16 timeseries and a 100x speedup in common NASA queries. The piece describes this as how Earthmover's Arraylake works, built on Icechunk and Zarr, and notes the same recommendations apply to humans and AI alike.

  2. Hacker News AI64

    Nvidia Study: AI Agents Rarely Need Long Context Windows

    An NVIDIA preprint by Belcak, Molchanov and co-authors reports that Claude Opus 4.5 and GPT 5.6 Sol lost little to no accuracy when their context was compacted into short windows across 13 long-horizon and long-input benchmarks, with accuracy shifts of −3.7 to +5.6 points for Opus 4.5 and −3.3 to +4.4 points for GPT 5.6 Sol on long-horizon tasks. Serving a 256K window cost 3.6× to 7× more per output token than a 16K window on open-weight models measured with vLLM on H100 GPUs, while all tested open models lost SWE-bench Verified accuracy at the tightest setting, with Qwen 3 Coder Next collapsing from 53.0% to 3.6%. The authors note the headline savings rely on their own price model rather than real bills, results come from single runs without confidence intervals, and all authors work at NVIDIA.

  3. Hacker News AI50

    Twenty Percent of all 'ok' web responses AI gets are not okay

    A new paper estimates that 20.9% of pages returning HTTP status 200 to a static client contain none of the page's main content, instead delivering app shells, bot challenges, block pages, error or placeholder pages, and partial content. Testing 10,307 web addresses over three days, the study found apparent citation drift rose from 5.8% to 21.9% when all 200 responses were accepted as successful retrievals, and recommends that fetching tools report what was actually retrieved rather than relying on status codes.

  4. Hacker News AI22

    Snowflake's AI_PARSE_DOCUMENT got an agentic mode

    A practitioner tested Snowflake's new AGENTIC mode for AI_PARSE_DOCUMENT on a SharePoint-fed Cortex Search pipeline, finding it costs a flat 8.18 credits per 1,000 pages versus 0.68 for OCR (about 12x) and only paid off on slide decks, where it returned chart, diagram, table and infographic summaries absent from OCR text. Two undocumented behaviors surfaced: image summaries live in each page's images array rather than the page content, and the mode parameter must be a literal, since a CASE expression silently returns NULL. The author settled on AGENTIC for PPTX and OCR for everything else, appending only the four useful image types to page text.

  5. Hacker News AI48

    Block training, keep answers? What 13,938 sites ask of AI bots

    Trakis Research analyzed robots.txt files from 13,938 readable sites in Common Crawl's September 2026 archive and found that of sites blocking OpenAI's GPTBot by name, 17.1% also block OAI-SearchBot and 25.9% block ChatGPT-User. The figure splits sharply by method: sites using Cloudflare's managed robots.txt rarely block answer bots (0.3% and 1.0%), while sites writing their own rules block them far more often (31.3% and 47.0%). Training bots are blocked about three times as often as answer bots, while Googlebot and Bingbot are almost never blocked.

  6. Hacker News AI36

    Automated grounding verification of AI statements about chest radiographs

    The paper presents BRACE, an automated framework that builds training examples from public radiology reports and Radiopaedia articles to check a statement about a patient's image against both the report and the cited evidence. Across 42 verifiers (0.1–284B parameters) on five public chest radiograph datasets, verifiers trained with BRACE accepted a mean of 0.239 of AI-written statements contradicted by the report at a threshold accepting 0.90 of supported examples, versus 0.830 for verifiers trained without contradicted examples, and accepted 0.169 of statements with an agreeing report but unsupporting citation, versus 0.830 for zero-shot LLMs. Two board-certified radiologists' independent labels agreed with the framework's labels (k > 0.910) nearly as closely as with each other (k = 0.952), and a recipe food-allergen instantiation accepted 0.493 of contradicted statements versus 0.898 for a lexical baseline.

  7. Hacker News AI58

    AI Can Read Neural Network Weights. I Wasn't Ready for That

    The author describes a study in which frontier models read neural network weights presented as plain text and reason about the behavior they encode, such as identifying a Pac-Man-playing network from its architecture and weights alone. He argues this is surprising because LLMs have no native channel for weights and no training data of this kind, and suggests it may reflect extreme generalization from program tracing and reinforcement learning. He notes that GPT-6 Astra, Opus 5.5 and Muse Spark 1.3 all identified Pac-Man, with Astra reporting 93% confidence.

  8. Hacker News AI36

    Show HN: Export your Character.AI chats in minutes – open-source, runs locally

    An open-source browser extension saves your Character.AI chats, characters, personas and memory boxes to a single local JSON file (plus optional plain text) in minutes, versus the official export's 3 to 4 weeks. It uses only activeTab and scripting permissions, reads without modifying anything, talks only to Character.AI with your signed-in session, and includes no analytics, servers or remote code. A free open-source viewer at amavie.ai/character-ai-export reads the file in your browser; the extension is MIT licensed and not affiliated with Character.AI.

  9. Hacker News AI26

    The Little Book of Generative AI Foundations: An Intuitive Mathematical Primer

    This revised 195-page preprint offers a compact, derivation-oriented introduction to the mathematical foundations of generative AI, tracing a coherent path from PCA and probabilistic PCA through variational autoencoders, diffusion models, normalising flows, autoregressive factorisations, GANs, Wasserstein GANs, and energy-based models. It aims to make the structure of generative modelling more accessible while retaining the mathematics needed to understand how these models are derived and related, and is intended for mathematically curious researchers, practitioners, and students. The version adds visual explanations throughout and welcomes comments and corrections.

  10. Fireworks AI Blog58

    Fireworks unveils new products at Forge to help companies build specialized intelligence

    At its first annual Forge conference, Fireworks launched products aimed at helping companies build AI around their own data and tasks, including Ultra-fast inference on NVIDIA Blackwell GPUs reaching up to 1,500 tokens per second, a half-price Background Mode, and multi-day prompt caching. It also introduced Fireworks Nexus for bringing open models into existing coding and coworking tools, No Eval and Specialized Model Inference for training and serving custom models, and Fireworks Sovereign, now in private preview, which runs its inference engine inside customers' own cloud or data centers. Fireworks says it already manages over 25,000 GPUs in customer environments.

  11. Databricks Blog34

    Declare and optimize with UNIQUE constraints on Databricks

    Databricks introduces UNIQUE constraints in Public Preview for Unity Catalog Delta tables in Databricks SQL and Runtime 18.2+, allowing tables to record multiple uniqueness guarantees beyond a single non-nullable primary key, with nullable columns and foreign key references supported. Constraints are informational and unenforced by default; after validation, marking them RELY lets the query optimizer eliminate eligible joins and DISTINCT aggregations, with benchmarks showing 7.9x and 12.0x speedups on TPC-DS SF10000.

  13. Ars Technica · AI58

    Ukraine’s drone strikes spur hasty Russian plans for underground data centers

    Ukrainian drone strikes have damaged Russian data centers housing AI supercomputers and disrupted cloud services, prompting Russian tech companies to scramble to harden and hide facilities underground. The Kremlin is requiring data center owners to reserve computing capacity for emergency state needs and has authorized possible use of foreign data centers. State-owned telecom Rostelecom has raised moving some data centers underground in talks with Norilsk, according to Vedomosti, and a regional governor proposed building data centers in open-pit coal mines with parks above as camouflage.

  14. TechCrunch · AI58

    Hiding inside Palantir’s skyrocketing quarter is a foreboding signal

    Palantir reported third-quarter revenue of $2.31 billion, up 96% year over year and above analysts' $2.18 billion expectation, with adjusted earnings per share of $0.44 versus $0.41 expected, and U.S. commercial revenue surpassed U.S. government revenue for the first time. However, U.S. enterprise total contract value fell to $1.46 billion from a record $2.13 billion the prior quarter, and CEO Alex Karp said enterprises are signing shorter-duration contracts that Palantir aims to expand later. Palantir raised its full-year U.S. commercial revenue guidance to more than $3.559 billion, at least 143% growth.

  15. SemiAnalysis62

    Elon Is Doing It Again: SpaceX Will Be The World’s Biggest AI Cloud

    SemiAnalysis reports that SpaceX now has 3.6GW of datacenter capacity under construction across multiple sites and has raised its 2027 estimate to roughly 13GW, which would exceed all of Google's datacenter capacity built from 1999 to 2026. The report says SpaceX may be the world's fastest datacenter builder, citing Colossus 2 build times of 91 days for 210MW and a planned ~4-month greenfield build at a Mississippi site, plus a datacenter likely planned at the TeraFab chip fab site in Texas. It also tracks over 8GW of likely behind-the-meter power, a ~6GWh BESS deployment at Colossus 2, and a 500kV substation connecting Macrohard (700MW) and MiniHard (450MW) to TVA service.

  16. Hacker News AI48

    The Largest Bitcoin Miners Are Quietly Becoming AI Data Centers

    Bitcoin miners have signed more than $70 billion in AI and high-performance computing contracts, with deals such as TeraWulf–Anthropic, CleanSpark–Meta, CoreScientific–AMD and Riot Platforms–Anthropic announced between Q2 and Q4, and analysts expect the industry to earn as much as 70 percent of its revenue from AI by the end of 2026. AI contracts pay more per electron and offer multi-year revenue certainty, while post-halving mining margins have fallen to multi-year lows. Miners still face an approximately $50 billion near-term funding gap and have so far delivered only about 25 percent of the AI and HPC capacity leased to customers.

  17. The Decoder58

    Rapid LLM development is outrunning the research meant to explain it, study finds

    Stanford researchers Owen Queen and James Zou tested 50 widely cited claims about large language models against 20 models and found 44 percent have already "expired," with an average lifespan of 15.7 months among the 22 findings with an estimated expiration date. Claims about what models can't do expired fastest, while findings on preferences, sycophancy and evaluator behavior proved longer-lived, and results varied widely across models even within the same release window. The researchers argue empirical LLM findings should be presented as observations about specific models at a specific point, with model, version, date, prompts and analysis code documented for re-testing.

  18. Hacker News AI72

    NYT demands Brave shut down its paywall blocker and end deals with AI firms

    The New York Times and other newspaper companies sued Brave Software, alleging the browser maker illegally bypasses paywalls and sells stolen news content to AI companies including Anthropic, Perplexity, Mistral, xAI, and Cohere. The suit, filed in the US District Court for the Southern District of New York, accuses Brave of direct, contributory, and vicarious copyright infringement and of violating the DMCA by circumventing paywall access controls. It also claims Brave's browser blocks ads on The Times's site and shows its own instead, depriving the publisher of digital advertising revenue.

  20. Hacker News AI64

    AI tool can predict cancer survival by a single patient selfie

    Researchers at Mass General Brigham developed an AI system, trained on more than 40 million facial photographs, that estimates biological age and predicts survival outcomes for cancer patients from a single selfie, with details published in Lancet Digital Health. The two tools, FAHR-FaceAge and FAHR-FaceSurvival, rely on different facial regions, and among radiotherapy patients in the US those who looked five or more years older than their actual age had a 21% higher risk of dying during the average four-year follow-up, while those who looked five or more years younger had a 17% lower risk. Doctors cautioned that more research is needed before the tool could be widely used.

  21. Tomasz Tunguz67

    The Financial Engineering Essential for AI in 2027

    Short-term GPU rental prices have risen from $3 to $13 per hour, pushing Anthropic's blended cost per GPU-equivalent hour from $3.10 to $5.65, according to the updated Anthropic S-1/A. With data center delays, idle GPUs awaiting power, and multi-year transformer and turbine lead times, labs are responding through model efficiency gains, tighter subscription allowances, and multi-year enterprise commitments. The piece argues that in 2027 the winners will be labs that match the duration of customer contracts to the duration of their compute.

  22. Hacker News AI40

    Counted "AI" on 37 website builder homepages, 2019 to now

    An analysis of Wayback Machine captures of 37 website, landing page and scrollytelling builder homepages found 570 mentions of "AI" today versus 16 in July 2022, with IONOS, Elementor and Squarespace leading at 76, 36 and 34 mentions. The piece notes that older selling terms like "templates" and "drag and drop" persisted rather than fading, and that "agent" went from zero mentions in July 2022 to 99 now. The underlying CSV is released under CC BY 4.0.

  23. Hugging Face Blog55

    StateBench: Testing how language models track state changes over long documents

    Researchers introduce StateBench, a benchmark with four difficulty levels of long-context state tracking, from immutable assignment to general finite problems involving assignment by reference. In tests, open pretrained models and frontier models (Gemini 3.8 Flash, Claude Opus 5.5, GPT-6 Sol and Luna) degraded on harder levels as inputs grew, while 275M-parameter Gated DeltaNet and hybrid models trained directly on the tasks generalized far beyond training length, though transformers never learned the hardest level. The work is released with a tech report, data, and code, and notes open questions such as fixed variable counts and the exact hierarchy placement of constrained Gated DeltaNet variants.