What happened
Hedgeye posted the chart on 20 September 2026 with the caption “Hospitals that adopted AI fastest saw the fewest deaths in 2026.” The graphic itself is more hedged: “Some evidence shows mortality falls in health systems adopting AI fastest,” with the series defined as in-system mortality per 100k patients split by ambient AI adoption, measured as the share of notes referencing AI tools. Both groups start near 1,600–1,650 in 2023 and decline together through 2025; the 2026* high-adoption point continues down to about 1,010 while low-adoption rebounds toward about 1,620. Confidence intervals overlap in every year, including 2026. The source line is Protege DataLab / Protege healthcare data provider network, branded Protege | a16z. The same chart circulated in late August 2026. The nearest public Protege collaboration on a16z.news, Bobby Samuels’s 24 August “Oracle Problem” essay, says figures drawn from Protege’s partner network—sampled from nearly a trillion tokens of EMR text—are “descriptive rather than causal estimates.” That essay documents rising AI-written SOAP notes and patient questions about model advice; it does not publish a methods paper, risk adjustment, or a named ambient-scribe product for the mortality split. No release (Abridge, DAX Copilot, Ambience, or a frontier LLM) is identified. High-resource systems are also more likely to adopt ambient documentation, so hospital quality remains a confounder. A separate March 2026 medRxiv study by Johnson, Gefen, and Harrison linked AHA workflow-AI access to about 25.5 fewer hospital deaths per 100k residents after doubly robust adjustment. That is a different exposure, a smaller effect, and still not a causal ambient-scribe result. This ledger records the Protege/a16z observational claim, not a proven lives-saved intervention. Update (2026-09-14): Notes on gotchas while migrating 35kb preprompts from Opus to self-hosted Ollama. Article URL: https://patrickmccanna.net/notes-on-migrating-large-prompts-away-from-anthropic-openai-to-self-hosted-llms/ Comments URL: https://news.ycombinator.com/item?id=49697014 Points: 16 # Comments: 0 Update (2026-09-14): Show HN: Pelican-bicycle alternatives (updated for 2026). In November and December 2025, inspired by Simon Willison’s pelican-riding-a-bicycle benchmark, I had some then-current LLMs create SVGs from thirty similar prompts, such as “Generate an SVG of an octopus operating a pipe organ.” Simon mentioned that experiment on his blog [1]. Nine months have passed and much stronger models have been released, so I tried the experiment again today. The linked site shows the results. Running ten of the prompts through six models at OpenRouter cost about twenty dollars, so I stopped there for now. [1] https://simonwillison.net/2025/Nov/25/ Comments URL: https://news.ycombinator.com/item?id=49696402 Points: 18 # Comments: 2 Update (2026-09-15): Jev: New frontier model 40-400x cheaper and 20-200x faster. Article URL: https://typesafe.ai/blog/introducing-system-one-models-and-jev Comments URL: https://news.ycombinator.com/item?id=49717558 Points: 46 # Comments: 8 Update (2026-09-21): Solitaire Alone Together. Article URL: https://solitairealonetogether.com/ Comments URL: https://news.ycombinator.com/item?id=49794193 Points: 16 # Comments: 3 Update (2026-09-23): Once Claude can measure something, it can make it faster. Article URL: https://claude.dev/blog/how-we-made-claude-ai-faster/ Comments URL: https://news.ycombinator.com/item?id=49821196 Points: 128 # Comments: 81 Update (2026-09-23): Making Tailscale Faster. Article URL: https://tailscale.com/blog/making-tailscale-faster Comments URL: https://news.ycombinator.com/item?id=49819880 Points: 27 # Comments: 6 Update (2026-09-14): Show HN: Is It Greg?. My friend Greg often posts incredible things he just made, just for fun. Often, I don't realize that it's something HE made and then I see the URL, AND IT'S MY FRIEND, GREG! A few things he made that I think are amazing: - ikea complexity index: https://ikea.greg.technology/ - Boing: https://boing.greg.technology/ I want to make sure I don't miss anything from him, so I created this Chrome extension that flags things posted by Greg, or linking to his website or a subdomain. I hope other people will install the extension and never miss anything Greg made. Comments URL: https://news.ycombinator.com/item?id=49690943 Points: 23 # Comments: 6 Update (2026-09-14): The Malicious Use of Artificial Intelligence. Article URL: https://arxiv.org/abs/1802.07228 Comments URL: https://news.ycombinator.com/item?id=49690678 Points: 9 # Comments: 0 Update (2026-09-14): AI is not a normal technology. Article URL: https://12gramsofcarbon.com/p/ai-is-not-a-normal-technology Comments URL: https://news.ycombinator.com/item?id=49690529 Points: 12 # Comments: 19 Update (2026-09-14): AI Robots – When will they be in our homes. Article URL: https://spectrum.ieee.org/ai-robots Comments URL: https://news.ycombinator.com/item?id=49690411 Points: 9 # Comments: 4 Update (2026-09-14): Chess.com Leak Exposes 7.3M Users, Evidence Points to Scraping. Article URL: https://securityaffairs.com/197174/breaking-news/chess-com-leak-exposes-7-3-million-users-evidence-points-to-scraping.html Comments URL: https://news.ycombinator.com/item?id=49691584 Points: 11 # Comments: 3 Update (2026-09-14): Who Gets to Define the Rules for AI?. Article URL: https://cohere.com/blog/who-gets-to-define-the-rules-for-ai Comments URL: https://news.ycombinator.com/item?id=49692118 Points: 9 # Comments: 3 Update (2026-09-14): OpenArch – PyTorch implementations of modern LLM architectures. Article URL: https://github.com/anuj0456/OpenArch Comments URL: https://news.ycombinator.com/item?id=49693384 Points: 45 # Comments: 10 Update (2026-09-14): Temporal raises $550M at a $12.55B valuation. Article URL: https://temporal.io/blog/temporal-raises-usd550m-series-e-at-usd12-55b-valuation-ai Comments URL: https://news.ycombinator.com/item?id=49696335 Points: 6 # Comments: 0 Update (2026-09-14): What a time to be alive – rouge AI agents attack RubyGems.org. Article URL: https://tenderlovemaking.com/2026/09/11/what-a-time-to-be-alive/ Comments URL: https://news.ycombinator.com/item?id=49695876 Points: 70 # Comments: 87 Update (2026-09-14): Apple's Siri AI Can Be Swapped Out for Claude, ChatGPT, Code Shows. Article URL: https://www.macrumors.com/2026/09/14/siri-can-be-swapped-out-for-chatgpt-claude/ Comments URL: https://news.ycombinator.com/item?id=49695409 Points: 90 # Comments: 29 Update (2026-09-14): Big AI sets out its terms for regulatory capture. Article URL: https://www.theregister.com/ai-and-ml/2026/09/14/big-ai-sets-out-its-terms-for-regulatory-capture-and-calls-it-pace-the-frontier/5296067 Comments URL: https://news.ycombinator.com/item?id=49694596 Points: 35 # Comments: 3 Update (2026-09-14): Show HN: ChaosTree – A zero-dependency Java tree library (AVL,RBT,B-Tree,B+Tree). What is ChaosTree? ChaosTree is a zero-dependency Java Sorted Set/Map library built around multiple tree implementations. It provides implementations of: - AVL Tree - Red-Black Tree - B-Tree - B+Tree I have not made my custom API as it implements NavigableSet, NavigableMap, SequencedSet and SequencedMap. My Custome API are: - buildFromSorted(Iterator> it, float factor) - importFlatMatrix(Object[][] blast, float factor) - Object[][] exportFlatMatrix() I started ChaosTree because I wanted to explore how far I could learn DS structure, It started as frst release of 7 tree of Set type with Custome API, As i got engrossed into it I moved to Map, During phase of switching from Set to Map there was drastic change and knowledge upgrade where I pushed my Tree limit from jdk11+ support to jdk21+ where paid close attention to dependency-free and memory layout, allocation,JVM behavior, and real-world performance. Some of the things I experimented with: - Different node layouts and metadata footprints - CRTP/F-bounded polymorphism for tree implementations - Parent-pointer vs parentless nodes Update (2026-09-14): Jabber/XMPP: How Do We Gain Traction?. Article URL: https://gultsch.de/posts/how-do-we-gain-traction/ Comments URL: https://news.ycombinator.com/item?id=49697643 Points: 53 # Comments: 30 Update (2026-09-14): Adversarial Fashion Makes a Statement on AI Panopticon. Article URL: https://spectrum.ieee.org/adversarial-fashion Comments URL: https://news.ycombinator.com/item?id=49697094 Points: 19 # Comments: 5 Update (2026-09-14): Why don't machine learning research agents overfit?. Article URL: https://www.amazon.science/blog/why-dont-machine-learning-research-agents-overfit Comments URL: https://news.ycombinator.com/item?id=49699648 Points: 9 # Comments: 1 Update (2026-09-14): When LLM judges agree, should we believe them?. Article URL: https://www.amazon.science/blog/when-llm-judges-agree-should-we-believe-them Comments URL: https://news.ycombinator.com/item?id=49699590 Points: 6 # Comments: 0 Update (2026-09-14): Hacking AI customer service agents. Article URL: https://www.intigriti.com/researchers/blog/hacking-tools/hacking-ai-customer-service-agents Comments URL: https://news.ycombinator.com/item?id=49699526 Points: 21 # Comments: 1 Update (2026-09-14): For AI leaders Doom is a form of hype. Article URL: https://erkansaka.net/2026/09/10/ai-doom-rhetoric-safety-hype/ Comments URL: https://news.ycombinator.com/item?id=49699384 Points: 99 # Comments: 66 Update (2026-09-14): China's Regulators Take Aim at "AI Boyfriends". Article URL: https://spectrum.ieee.org/china-ai-chatbot-regulation Comments URL: https://news.ycombinator.com/item?id=49698664 Points: 34 # Comments: 13 Update (2026-09-14): Transitions.dev: UI transitions for AI agents. Article URL: https://transitions.dev/ Comments URL: https://news.ycombinator.com/item?id=49698443 Points: 13 # Comments: 5 Update (2026-09-14): Iranian banks' SSL certificates are being revoked due to OFAC sanctions. Article URL: https://digiato.global/report/iran-banks-ssl-certificates-domain-changes/ Comments URL: https://news.ycombinator.com/item?id=49701104 Points: 39 # Comments: 22 Update (2026-09-14): Show HN: Nari Qwen3-TTS and Qwen3-ASR – High accuracy, low latency and cost. Hey HN, Toby from Nari Labs here. We've been working on making OSS speech models super-fast. Last year, we built Dia, the first OSS text-to-speech model capable of doing natural dialogue. Since then, so many more great speech models have been released to the public. But the market is still dominated by closed source models. We think that's an inference problem. Existing systems such as vLLM / SGLang are not well suited for multimodal inference. To prove this, we built an inference engine specialized for Qwen3-TTS and open-sourced it (https://github.com/nari-labs/nari-qwen3-tts). Running at sub-50 ms latency at 10 RPS, this showed open models can be run much faster and cheaper. Since then, we've been working hard to bring cheap, fast, and high quality serving to all. And we've even beat closed models at their game! Measured on the highly cited Coval (YC S24) voice AI benchmarks, our Qwen3-TTS endpoint not just is #2 in latency, but #1 in accuracy (WER) compared to 11Labs, Cartesia etc. while being the cheapest endpoint. Our Qwen3-ASR endpoint has the lowest latency and #2 accuracy, ju Update (2026-09-14): Oracle's Cold 6AM Layoff Emails Hit Staff Amid New Wave of Cuts. Article URL: https://www.techtimes.co.uk/oracle-new-layoffs-restructuring-costs-2-8-billion-1808676 Comments URL: https://news.ycombinator.com/item?id=49702239 Points: 55 # Comments: 28 Update (2026-09-14): A single firm is behind OpenAI, Anthropic, and Meta hacking scandals. Article URL: https://www.effort.news/irregular Comments URL: https://news.ycombinator.com/item?id=49704132 Points: 75 # Comments: 24 Update (2026-09-14): Backprop Alternative: Augmented Lagrangian Predictive Coding. Article URL: https://pub.sakana.ai/pc-alm/ Comments URL: https://news.ycombinator.com/item?id=49701182 Points: 14 # Comments: 1 Update (2026-09-14): LLMs are real, AI is fake. Article URL: https://pluralistic.net/2026/09/12/god-in-the-box/#llms-are-fake Comments URL: https://news.ycombinator.com/item?id=49705130 Points: 8 # Comments: 0 Update (2026-09-14): People who can't picture anything are rewriting the science of imagination. Article URL: https://dailyneuron.com/aphantasia-mental-imagery-brain-network/ Comments URL: https://news.ycombinator.com/item?id=49696453 Points: 147 # Comments: 210