Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI – Page 103

Bananas, cups and peelers: Robots learn how to handle curved objects like fruits and tools

It does not take much to confuse some robots. A machine might be great at handling a simple object like a box, yet when it tries to work with a more irregular shape like a banana, it often fails.

But help is at hand. Researchers from the Swiss Federal Technology Institute of Lausanne (EPFL) and Idiap Research Institute have developed a new approach that lets robots more reliably manipulate a variety of different shapes by teaching them to follow the unique geometry of any object they encounter.

Their work is detailed in a paper published in the journal Science Robotics.

Machine learning offers faster, more reliable analysis of Fermi surfaces in search of spintronic materials

The search for next-generation electronic materials often starts with studying the Fermi surface, which serves as a map of a material’s electronic structure. Its shape varies with crystal structure, composition, and electronic band arrangement, directly impacting properties such as carrier density, magnetic behavior, and spin polarization. This makes it a crucial tool for understanding and engineering new materials.

The Fermi surface of a material is determined experimentally using techniques such as angle-resolved photoemission spectroscopy (ARPES). However, interpreting ARPES data requires specialized expertise, and the measurements themselves are often susceptible to noise. As experiments produce larger amounts of data, carefully reviewing every image by hand becomes time-consuming and inefficient.

Better volcano eruption predictions on Earth—and Venus—thanks to Mauna Loa study

When Mauna Loa erupted in 2022, the largest lava flow headed on a path headed directly toward Daniel K. Inouye State Highway 200, also known as Saddle Road, a critical route that carries many residents from their homes on one side to their jobs on the other.

No one could accurately predict whether the lava would continue to flow and eventually block the highway, or stop short, sparing the road.

However, when the volcano next erupts scientists will be better able to monitor the eruption in real time and make more accurate predictions about where the lava will flow and when the volcano might erupt. These advances are thanks to the availability of satellite data from public and private sources as well as machine learning algorithms developed at Pitt with help from a colleague in Italy, as highlighted in a recent publication in the Journal of Volcanology and Geothermal Research.

Neural network speeds tuning of attosecond light pulses for physics experiments

Researchers from Skoltech and the Shanghai Institute of Optics and Fine Mechanics have developed an approach that helps optimize the parameters of a laser-plasma source of attosecond pulses—ultrashort flashes of light used in physics experiments. Instead of relying on a large number of time-consuming calculations, the team trained a neural network to quickly identify promising settings and thereby speed up the optimization of the sophisticated laboratory equipment.

The results were published in Communications in Nonlinear Science and Numerical Simulation.

Attosecond pulse sources are used as research tools. They are applied in ultrafast spectroscopy, studies of magnetic materials, chiral molecules, and electron dynamics in matter. The goal of this work is to make it faster to tune a light source with the required properties for such experiments.

DuctGPT demonstrates how AI can accelerate discovery of next-generation fusion materials

Scientists at Ames National Laboratory developed a new artificial intelligence (AI) tool that accelerates discovery of materials needed for next-generation fusion energy systems. The tool, DuctGPT, combines advanced AI with physics-based modeling to help researchers predict materials with the appropriate properties to function in the extreme conditions inside of fusion reactors.

This research is discussed in “DuctGPT: A Generative Transformer for Forward Screening of Ductile Refractory Multi-Principal Element Alloys,” published in Acta Materialia.

The challenge is to rapidly explore a wide range of potential alloy compositions that can maintain high-temperature strength, while retaining the ductility necessary for manufacturing the materials.

Scientists Teach AI To Think Like a Professional Chemist

Researchers have developed a framework that interprets chemical strategy as language, opening a new path for AI-assisted discovery. Designing molecules is one of the most difficult tasks in chemistry. Whether creating new medicines or advanced materials, each compound must be built through a care

The Why Is a Discipline: Goodhart’s Law and AI

A reader asked me a question this week that I have been thinking about ever since.

She did not ask whether AI could malfunction. She did not ask whether bad actors could misuse it. She asked something sharper:

Can a system produce bad outcomes systematically, even when intent is good, and nothing is broken?

The answer is yes. And it is the most dangerous category of bad outcome, because nobody is at fault and nothing is broken.

We have all the evidence we need. Amazon ran into it. YouTube ran into it. Hospitals are running into it now. AI labs are about to run into it at a planetary scale. And almost nobody is talking about why.

A 1975 economic principle explains it cleanly. A reader’s question forced me to refine an argument I have been making for years.

New essay: [ https://www.singularityweblog.com/goodharts-law-ai/](https://www.singularityweblog.com/goodharts-law-ai/)

Continue reading “The Why Is a Discipline: Goodhart’s Law and AI” | >
/* */