AI & Robotics News and Discussions

[firestar464]

man what an uninformed study on LLMs

https://www.bmj.com/content/387/bmj-202 ... -responses

[firestar464]

AI 'brain decoder' can read a person's thoughts with just a quick brain scan and almost no training

https://www.livescience.com/health/mind ... o-training

[wjfox]

AI cameras that spot phone offences to be trialled

17 February 2025

State-of-the art cameras designed to identify drivers not wearing seatbelts and using mobile phones will be trialled by police in Essex.

The high-definition AI (artificial technology) has been used by other forces across England.

Motorists can be sent warning letters, fined and prosecuted as a result of the cameras being used.

The technology is due to be deployed by Essex Police from April and the force's head of roads policing, Adam Pipe, said the results he had seen were "phenomenal".

"The picture quality is brilliant and they are detecting offence numbers that we would simply never detect by an officer being present at the roadside" said Mr Pipe, speaking to BBC Essex.

https://www.bbc.co.uk/news/articles/ckgxp9g5njdo



Credit: Image source, Vision Zero South West

[firestar464]

[wjfox]

[weatheriscool]

Helix: The hive-minded, fully autonomous VLA humanoid robot
By Joe Salas
February 21, 2025

https://newatlas.com/robotics/helix-vla ... -02-robot/

Only weeks after Figure.ai announced ending its collaboration deal with OpenAI, the Silicon Valley startup has announced Helix – a commercial-ready, AI "hive-mind" humanoid robot that can do almost anything you tell it to.

Figure has made headlines in the past with its Figure 01 humanoid robot. The company is now on version 02 of its premiere robot, however, it's gotten more than just a few design changes: it's gotten an entirely new AI brain called Helix VLA.

It's not just any ordinary AI either. Helix is the very first of its kind to be put into a humanoid robot. It's a generalist Vision-Language-Action model. The keyword being "generalist." It can see the world around it, understand natural language, interact with the real world, and it can learn anything.

[Yuli Ban]

[firestar464]

Can AI help beat poverty? Researchers test ways to aid the poorest people

https://www.nature.com/articles/d41586-025-00565-7

[weatheriscool]

Figure's humanoid robots will take on your household chores this year
By Abhimanyu Ghoshal
February 27, 2025

Figure is leveling up its humanoid robot business far more rapidly than any other robotics company I can think of right now. CEO Brett Adcock just announced it plans to begin alpha testing its robots in the home sometime in 2025 – that's two years ahead of schedule.

Adcock says the update to Figure's timeline is thanks to swift advancements in its Helix AI, which we heard about just last week. The Bay Area startup noted this AI brain was the first of its kind – a generalist Vision-Language-Action model, to be precise – to be integrated into a humanoid robot. The idea is to enable Figure bots to see what's happening around them, understand natural language, interact with the real world, and learn to do just about anything.

https://newatlas.com/robotics/figures-h ... tt-adcock/

[raklian]

First AI-written paper passes human peer review, accepted for scientific publication

Summary

We are proud to announce that a paper produced by The AI Scientist passed the peer-review process at a workshop in a top machine learning conference. To our knowledge, this is the first fully AI-generated paper that has passed the same peer-review process that human scientists go through.1

The paper was generated by an improved version of the original AI Scientist, called The AI Scientist-v2. We will be sharing the full details of The AI Scientist-v2 in an upcoming release. This paper was submitted to an ICLR 2025 workshop that agreed to work with our team to conduct an experiment to double-blind review AI-generated manuscripts. We selected this workshop because of its broader scope, challenging researchers (and our AI Scientist) to tackle diverse research topics that address practical limitations of deep learning. The workshop is hosted at ICLR, one of three premier conferences in machine learning and artificial intelligence research, along with NeurIPS and ICML.

We conducted this experiment with the full cooperation of both the ICLR leadership and the organizers of this ICLR workshop. We thank all of them for supporting this research into how AI-generated papers fare in peer-review. Furthermore, we also received an institutional review board (IRB) approval for this research from the University of British Columbia. Lastly, we plan to give a talk at the ICLR workshop to share our experiences and particularly the challenges with the AI Scientist project.

We proudly collaborated with the University of British Columbia and the University of Oxford on this exciting project.

https://sakana.ai/ai-scientist-first-publication/

[firestar464]

wow

[firestar464]

TxAgent: An AI agent for therapeutic reasoning across a universe of tools

https://zitniklab.hms.harvard.edu/TxAgent/

[firestar464]

Glimpses of AI Progress

https://www.pathwaysai.org/p/glimpses-of-ai-progess

[spryfusion]

[wjfox]

Browser Use, the tool making it easier for AI ‘agents’ to navigate websites, raises $17M

March 23, 2025

We may not have an agreed-upon definition of AI “agent” yet, but a multitude of startups want to create “agentic” tools to automate various tasks online. One such firm, Browser Use, has attracted a ton of interest from developers and investors thanks to its solution that makes websites more “readable” for AI agents.

[...]

Browser Use essentially breaks down the buttons and elements of a website into a more digestible, “text-like” format for agents. This helps the agents understand the different options and make decisions autonomously.

“A lot of agents rely on vision-based systems and try and navigate websites through screenshots, and in [the] process, things break,” Müller said. “We convert [websites] into something agents can understand. This approach means we can run the same tasks again and again at a cheaper cost.”

There’s an increasing number of AI companies that want to make their agents interact with websites more gracefully, and Müller thinks Browser Use can become a “fundamental layer” serving this need. He added that more than 20 companies in the current Y Combinator Winter batch used Browser Use for their own requirements.

“There are companies coming to us and saying, ‘What can we do to make it easier for agents to navigate our website?’” Müller said. “There are sites — for example, LinkedIn — that change the way the website works all the time, so agents often fail on sites like those.”

https://techcrunch.com/2025/03/23/brows ... aises-17m/

[weatheriscool]

[firestar464]

Who else expects the robot to be teleoperated again

[firestar464]

wow

Gemini 2.5 for comparison:

https://aistudio.google.com/app/prompts ... :%7B%7D%7D

I'm honestly not sure where "Mila" and "Kai" come from.

[weatheriscool]

[weatheriscool]

Record-breaking tiny robot offloads electronics to fly by magnetism
By Ben Coxworth
March 31, 2025
https://newatlas.com/robotics/worlds-sm ... tic-field/

Scientists have created what they say is the world's smallest untethered flying robot, by taking a unique approach to its design. To minimize size and weight, they've moved the bot's power and control systems out of its sub-centimeter-wide body.

Measuring just 9.4 mm in width and tipping the scales at 21 mg, the robot is being developed by Prof. Liwei Lin and colleagues at the University of California, Berkeley.

It mimics the flight capabilities of the bumblebee. Like that insect, it can hover in place, move both vertically and horizontally, and hit small targets. Its descendants could one day perform tasks such as pollinating crops, or exploring spaces too small for ordinary drones to access.

The bot's 3D-printed polymer body consists of a four-bladed horizontal propeller, encircled by a "balance ring." Protruding up from the center of the propeller is a small vertical ring that holds two puck-shaped neodymium permanent magnets – each one is 1 mm wide by 0.5 mm thick.