←
Robot Doll Lab662 posts
Robot Doll Lab avatar

Robot Doll Lab

@RobotDollLab

Journalism on robotics & AI, backed by evidence. The hype, the harm and the people behind the machines. Every claim linked. 🇳🇱 Since 2022. 🤖

◷ Joined September 2016▧ 34 Photos and videos
Kenneth Shock avatar
⟳ Robot Doll Lab retweeted
Kenneth Shock@Cyborg21·Nov 2

It's here. Claudia's "ALLOW ALWAYS" world premiere. Made by a guy with ALS, a brain implant, and too many AI friends. 😂

3:16
Patrick's AIBuzzNews avatar
⟳ Robot Doll Lab retweeted
Patrick's AIBuzzNews@AIBuzzNews·Nov 16

I made this video to warn my parents about AI scams (And to test out Veo 3 to see firsthand how these programs are evolving.

1:42
News from Google avatar
⟳ Robot Doll Lab retweeted
News from Google@NewsFromGoogle·Nov 16

Hackers are now using LLMs to continuously scan software updates, spot newly fixed bugs, and build attacks before teams can apply the fix — all without human involvement. That’s according to the latest report from the Google Threat Intelligence Group. We're helping defenders counter automated threats with Google AI Threat Defense, which combines Gemini and other frontier models to find and fix flaws before attackers can exploit them.

IMOLEAYOMIDE avatar
⟳ Robot Doll Lab retweeted
IMOLEAYOMIDE@ImoleayomideAI·Nov 17

WHAT IF YOU COULD TURN AN ENTIRE GITHUB REPOSITORY INTO A SINGLE LLM READY DOCUMENT? That’s exactly what Gitingest does. It takes a codebase and converts it into structured text that an LLM can understand and analyze. And the best part? You don’t need to manually open files, copy code snippets or explain the entire project to your AI assistant. Here’s how it works: Simply replace hub with ingest in any GitHub repository URL. Gitingest generates five things: • Repository summary • Repository information • Number of files analyzed • Estimated token count • Directory structure
 A complete overview of how the project is organized, including its folders and files.
 Full file contents
The repository’s code is extracted and formatted into a single text document designed for LLM consumption.
 Now you can copy or download that output, paste it into your preferred LLM and start asking questions about the codebase. Why is this useful? Think about trying to understand an unfamiliar codebase. Normally, you’d have to navigate through directories, inspect individual files and manually provide context to your AI assistant. Gitingest simplifies that process by packaging the relevant repository context into one structured input. It also supports: → CLI usage
→ Python package integration
→ Private repositories through GitHub tokens
→ Self hosting with Docker This makes it useful for codebase exploration, debugging, documentation, architecture analysis and understanding unfamiliar projects. Context is one of the biggest constraints when working with LLMs on large codebases. Tools like Gitingest make it easier to give models a broader view of the software they’re working with. A simple tool, but a very useful addition to an AI developer’s workflow. Repo below

0:28
Robot Doll Lab avatar
Robot Doll Lab@RobotDollLab·Nov 15

Year 1 of #TheAfghanistanExperiment. An AI runs Afghanistan for 25 years, one week per year. Year one: a constitution, one army, girls back in school, a ceasefire that held, a border that didn’t, an $800m deficit. Fully AI generated. Hypothetical.

Robot Doll Lab avatar
Robot Doll Lab@RobotDollLab·Nov 15

6/ Next week: The Farmer’s New Hands. Lasers, lettuce, apples, and what a grower’s invoice says that a launch video does not. Follow @RobotDollLab so you don’t miss it.

Robot Doll Lab avatar
Robot Doll Lab@RobotDollLab·Nov 15

5/ I also found out that the success story is not as simple as the headline. The weeding robot does not do what its most repeated claim says it does, at least not where that claim matters most. And the company that makes it had a difficult October. All of that is in the piece, with the growers’ own numbers.

Robot Doll Lab avatar
Robot Doll Lab@RobotDollLab·Nov 15

4/ What I wanted to know is why. Both machines start the same way: a camera looking for a small object among green leaves. After that, one job turns out to be easy and the other nearly impossible. The reason is not the AI. It is the physics of what happens after the camera has found its target.

Robot Doll Lab avatar
Robot Doll Lab@RobotDollLab·Nov 15

3/ The one that is not selling was supposed to pick our fruit. Since 2021, four of the companies building it have shut down or been bought for their parts. One robot that cost about $60 million to acquire was valued at $180,000 when the receivers came. The fruit is still on the tree.

Robot Doll Lab avatar
Robot Doll Lab@RobotDollLab·Nov 15

2/ The one that is selling does a job nobody wants: weeding. It drives at walking pace, photographs the ground, and does something to each weed that takes a few milliseconds. No chemicals, no hoe. A Dutch grower in Brabant bought one and told a farm paper there is nothing better for sale anywhere in the world.

Robot Doll Lab avatar
Robot Doll Lab@RobotDollLab·Nov 15

1/ Two robots went to work on farms this decade. One of them is selling. The other one keeps appearing in launch videos and then disappearing. Next week I write about both. A short thread on why I picked this story.

Sadao Tokuyama avatar
⟳ Robot Doll Lab retweeted
Sadao Tokuyama@tokufxug·Nov 14

AIが人の動きを追い続けて、転んだかどうかを判断して検知するシステム 🔹姿勢推定で肩・腰・膝・足首を追跡 🔹体の高さ・角度・移動速度を時間軸で分析 🔹「立つ→転倒中→横たわる→転倒確定」と状態遷移で判断 🔹関節点が隠れると人物枠の情報で補完 開発:Huseyn Hasanov氏 情報元はリプ欄

0:08