标签：#ann

"I pretty much, at that point in time, gave up on being an astronaut."

On this week’s episode of The Big Interview podcast, WIRED’s senior culture editor Manisha Krishnan talks to Gowanus about eschewing Big Tech, going outside, and rejection in the age of dating apps.

Yes, Sarah Wynn-Williams violated her deal to stay silent. But the power disparity bolsters the view that Meta is a heartless bully.

Today on Uncanny Valley, we unpack the political debacle unfolding in Maine surrounding the campaign of Democratic candidate Graham Platner.

Every system that does "semantic" anything — RAG pipelines, recommendation engines, image search, dedup — boils down to one operation: given this vector, find the closest ones out of millions. The vectors are embeddings, a few hundred to a couple thousand numbers each, and "closest" means closest in meaning. You'd assume the database either scans all of them (slow but correct) or uses some clever tree to jump straight to the answer. It does neither. Instead it deliberately settles for the approximately closest vectors — and that compromise is the entire reason vector search is fast enough to exist. Two algorithms do almost all the heavy lifting in practice, in pgvector, Qdrant, FAISS, and the rest: IVF and HNSW . Here's what they're actually doing under the hood, and how to choose between them. Why "exact" is off the table The natural objection is: why approximate? Just find the real nearest neighbor. In two or three dimensions you could — a k-d tree or similar structure prunes away big regions of space and finds the true closest point quickly. The trouble is that embeddings live in hundreds of dimensions, and high-dimensional space is deeply weird. It's called the curse of dimensionality . As dimensions grow, the distance to your nearest point and the distance to your farthest point drift toward being almost the same. Formally, the contrast (d_max − d_min) / d_min shrinks toward zero. When everything is roughly equidistant from everything else, a tree can't confidently say "skip this whole branch, it's too far" — the bounding regions all overlap, every branch looks plausible, and the search degrades into checking nearly everything. Exact indexes quietly collapse back into brute force. So we change the question. Instead of "prove you found the nearest," we ask "quickly find something very probably among the nearest." That's approximate nearest neighbor (ANN) search, and it swaps a guarantee for speed. The quality knob becomes recall : of the true top-k neighbors, wh

Manna is launching a U.S. operations and manufacturing facility in Tulsa, Oklahoma, that will eventually employ 1,000 people.

Livestreaming the channel through its app now starts at $5 per month.

Station F, the Paris-based startup hub founded by French billionaire Xavier Niel in 2017, is gearing up for a new edition of its selective acceleration F/ai program that will cement its role as a stepping stone for AI startups.

It's difficult to pinpoint the moment in my life where America started to lose the plot.

GitHub had 20,000+ secret scanning alerts across 15,000 repositories. Here's how we separated signal from noise, built remediation workflows, and reached inbox zero in nine months. The post How GitHub used secret scanning to reach inbox zero appeared first on The GitHub Blog .

A timely tale about a 50-year-old robotic arm...

We're excited to launch PaperQuire — a desktop app that turns plain Markdown into professional, print-ready PDFs. No cloud uploads, no accounts, no subscriptions required for personal use. Why We Built PaperQuire If you write in Markdown, you've probably hit this wall: your content looks great in your editor, but the moment you need to share it as a polished document — a proposal, a report, a spec — you're stuck copy-pasting into Word or fighting with LaTeX. We wanted something simpler. Write in Markdown, click Export, and get a document that looks like a designer made it. No extra steps, no cloud dependency, no learning curve. What PaperQuire Does Live preview — See your formatted document side-by-side as you type. What you see is what you'll get in the PDF. Professional templates — Choose from templates designed for technical docs, proposals, reports, and more. Every template supports custom branding: your logo, your colors, your fonts. Offline-first — Your documents never leave your machine. PaperQuire runs entirely on your desktop — macOS, Windows, and Linux. Plugin system — Extend PaperQuire with plugins for diagrams (Mermaid), math (KaTeX), syntax highlighting, and more. AI Assist — Bring your own API key and get writing suggestions, grammar fixes, and content generation right inside the editor. How It Works Open PaperQuire and start writing Markdown — or open an existing .md file Choose a template and customize your branding (logo, colors, fonts) Click Export to generate a polished PDF instantly Share your document with confidence The entire process takes seconds, not minutes. Free for Personal Use PaperQuire is free for personal use with no restrictions on the core features. The Pro plan adds advanced exports (DOCX, HTML), batch processing, and priority support for teams that need more. Get Started Download PaperQuire for your platform: macOS (Apple Silicon) Windows (x64) Linux (x86_64) Check out the documentation for a quick walkthrough, or just start writi

For decades, the senator has argued that concentrated wealth threatened American democracy. Now he’s betting that frustration with Big Tech, billionaires, and unchecked AI is reaching a tipping point.

Have you noticed how every demo of "an AI agent" looks impressive in the video and falls apart the moment you ask a sharper question? The agent confidently does the wrong thing. It forgets what it just decided. It tries to call a tool that doesn't exist. It loops forever rewriting the same file. It calmly tells you the deployment succeeded when it didn't. These aren't failures of the model. They're failures of the workflow around the model. Because that's all an agent really is: a software workflow where a language model can pick the next step and call tools. The "intelligence" sits in the prompt and the orchestration around it, not in some secret agent-flavoured fairy dust. Strip the word "agent" away and you've got five pieces of plumbing: planning, tool use, memory, constraints, verification. Every production-grade agent stands or falls on those five. This is a long walk through each one. Not the marketing version. The kind of detail you actually need before you ship something that talks to your database. The Loop You're Actually Building Before we touch any pillar individually, hold the whole loop in your head. A useful agent does roughly this on every turn: Read the goal (and whatever memory is relevant to it). Decide the next action: answer directly, call a tool, ask a clarifying question, or stop. If it called a tool, observe the tool's result and feed it back in. Update memory if anything is worth remembering. Check constraints: are we over budget, out of iterations, touching something off-limits? Verify the output before declaring success. Loop until done or stopped. That's it. Every framework (LangGraph, OpenAI Agents SDK, Claude Agent SDK, smolagents, whatever ships next month) is a different shape of the same loop with different defaults. agent-loop.ts async function runAgent ( goal : string , ctx : AgentContext ) { const state = ctx . startState ( goal ); for ( let step = 0 ; step < ctx . maxSteps ; step ++ ) { const decision = await ctx . model . decid

It's a stretch to think that the continent can build a top-tier model, but it has an advantage: Donald Trump.

Amazon-owned MGM Studios’ decision to drop the OpenAI movie is just part of AI and film industries becoming increasingly intertwined. On Uncanny Valley, we take a look at where this is all headed.

A monorepo can look like one repository, but security teams should treat it as many applications living under one roof. One repo may contain 10 frontend packages, 5 backend services, 3 shared utility libraries, 2 mobile apps, and one root lockfile that does not tell the full story by itself. Monorepo dependency security means scanning the root dependency graph, every workspace package, shared libraries, lockfiles, and generated SBOMs. If you scan only one file, you may miss the vulnerable package that ships in production. Why Monorepos Create Unique Vulnerability Challenges Monorepos centralize multiple packages, apps, services, and libraries inside one repository. This improves code sharing, dependency alignment, refactoring, CI caching, and cross-team collaboration. It also creates a security problem: one repository can contain many different dependency trees, owners, deployment targets, and risk profiles. A typical JavaScript or TypeScript monorepo may include apps/web , apps/admin , apps/api , packages/ui , packages/auth , packages/logger , and packages/config . Each package may have its own package.json . Some packages are deployed to production. Some are internal libraries. Some are build-only tools. Some are used by every app. A vulnerability in one package can affect one app, many apps, or the whole repo depending on how dependency relationships are structured. The biggest issue is shared code. If packages/auth depends on a vulnerable version of jsonwebtoken , every application that imports packages/auth may be affected. If packages/ui uses a vulnerable utility such as lodash , every frontend app that consumes that UI package may inherit the same risk. If a build tool dependency is compromised, the risk may appear during CI/CD rather than runtime. Real CVEs show why this matters. CVE-2021-23337 affected lodash through command injection in template handling. CVE-2022-31129 affected moment through inefficient parsing that could cause denial of service. CVE-202

President Trump probably can’t get rid of her yet, but FCC commissioner Anna Gomez still checks her email every day to see if he has. Until then, she wants to stand up for the First Amendment.

Everything from 1.4.1 through 1.4.9 happened inside a single function, standard_planner() . Building paths, costing them, searching for a join order, estimating cardinality from statistics: all of it runs inside that one function. Yet PostgreSQL does not call standard_planner() directly. It puts another function, planner() , one step ahead of it, and has planner() call standard_planner() . And planner() can be made to call some other function instead of standard_planner() . That replacement is what the planner hook enables. When pg_stat_statements measures per-query planning time, or pg_hint_plan rewrites a plan according to hints, it all goes through this hook. Let's look at how PostgreSQL provides a way to observe or change planning behavior without touching a single line of the core, and how external code plugs into it. All planner() does is check the hook The body of planner() is essentially this. if ( planner_hook ) result = ( * planner_hook ) ( parse , query_string , cursorOptions , boundParams ); else result = standard_planner ( parse , query_string , cursorOptions , boundParams ); planner_hook is a global function pointer. Its default value is NULL , in which case standard_planner() is called right away. A plain PostgreSQL build always takes this path: planner_hook is empty, so the incoming query goes straight to standard_planner() . The key here is the type of planner_hook . typedef PlannedStmt * ( * planner_hook_type ) ( Query * parse , const char * query_string , int cursorOptions , ParamListInfo boundParams ); This signature is identical, down to the character, to that of planner() and standard_planner() . It takes the same Query and returns the same PlannedStmt (the execution plan). So external code only has to write a planner function matching this type and store its address in planner_hook . Let's call this function, written by external code to register in planner_hook , a custom planner function. The moment its address is stored, every planning reque

On today’s Uncanny Valley, we dive into the dysfunction in Meta’s newly formed AI unit and why it’s been driving already-low employee morale even further into the ground.