×

Exciting News! Our Mobile App is Here!

Experience the best of Mindplex on the go! Download now for a seamless, fast, and intuitive experience.

Download Now

Archives: News

news custom post type.

AI for solar system boundary exploration

Researchers are studying how artificial intelligence (AI) can aid a Chinese mission to the solar system’s edge, SpaceNews reports. This mission faces tricky problems like unknown places, long distances, and slow communication with Earth.

A paper published in Journal of Deep Space Exploration in late 2024 says AI can help solve these issues.

The paper comes from experts at institutions including Beijing Institute of Technology, Shanghai Academy of Spaceflight Technology, and China’s Deep Space Exploration Laboratory.

The paper explains how AI can make spacecraft more independent. For example, AI can process data, which is sorting and handling information, right on the spacecraft. This cuts down on how much humans on Earth need to do. AI can also spot things on its own, like strange space events, and make decisions without waiting for Earth’s instructions. Plus, it can save power by using efficient computing.

Two AI-assisted Chinese spacecraft will travel 100 astronomical units

China wants to send two spacecraft to the heliosphere’s head and tail by 2049. The heliosphere is the bubble around the Sun made by solar wind. These spacecraft will travel 100 astronomical units, where one AU is the distance from Earth to the Sun. Later, they aim for 1,000 AU by the century’s end. They’ll use Jupiter flybys and might visit other planets or Kuiper Belt objects. Their goals include studying dust, space rays, and the solar system’s edge. They’ll carry tools like cameras, dust analyzers, and a magnetometer, which measures magnetic fields.

AI could clean data, removing mistakes before sending it home. It might also shrink data using autoencoders that keep only key details. This helps because communication takes a long time over huge distances. AI can also watch the spacecraft’s health, guessing when parts might break. It could steer the spacecraft too, adjusting paths on its own using reinforcement learning, a method where machines learn from their surroundings.

China has used AI in space before, like on the Chang’e-6 moon mission. NASA uses it too, such as on the Mars rover Perseverance. This paper shows AI’s growing role in space. Only a few missions, like Voyager and New Horizons, have reached the solar system’s edge. China’s AI plans could make its mission stand out.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

New encyclopedia of human gene functions unveiled

The Gene Ontology Consortium has released a detailed guide to over 20,000 human genes that make proteins. The Consortium is funded by the National Institutes of Health and includes researchers from the Keck School of Medicine at USC, the Swiss Institute of Bioinformatics, and other places.

The researchers built a website where anyone can search this guide.

The researchers used evolutionary modeling to look at how genes changed over time across humans and animals like mice and fish. A paper published in Nature describe this study.

The Gene Ontology is a knowledge base that’s been growing for over 25 years. Scientists use it a lot: over 30,000 studies yearly rely on it. The knowledge base assists “omics” studies that look at DNA, RNA, proteins, and other stuff in cells all at once. For example, a scientist might find genes acting differently in cancer cells. Checking thousands of papers for each gene’s role is too hard. So, scientists use this knowledge base instead.

This tool turns gene lists into clear biological stories. Sometimes, it even hints at new treatments. The latest update makes it stronger by adding evolutionary clues. It mixes human gene data with animal gene data. This fills gaps where human studies are missing.

The new resource covers 82% of protein-coding genes

Over 150 biologists worldwide helped build this resource. Since 1998, they’ve read over 175,000 papers. They sorted genes by their tasks, like cell division or immune defense, using a list of over 40,000 functions. The new resource, called the PAN-GO functionome, covers 82% of protein-coding genes. For the rest – about 3,600 genes – no data exists yet. According to the researchers, this shows where future research should go.

Scientists can now use this tool with computers and artificial intelligence (AI), to dig into data fast. Users can and should suggest updates via the website. This keeps the guide fresh and useful. The researchers are persuaded that, tough not perfect, this is the best map of human gene functions so far.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

Other interesting news for 02.28.25

This post collects links to other interesting news that we have considered but not included in our regular news this week.

The International Space Station is overly sterile; making it “dirtier” could improve astronaut health

AI unlocks the emotional language of animals

Why GPT can’t think like us

Illuminating the proton’s inner workings

New microscope can image, at once, the full 3D orientation & position of molecules in cells

A bioinspired robot can shapeshift to drive, roll, or swim through rough terrains

Studying locusts in virtual reality challenges models of collective behavior

Fiber computer allows apparel to run apps and “understand” the wearer

A new device enables remote tasting in virtual and augmented reality

Zero-shot AI for remote sensing: A new pipeline for automated image segmentation

Efficient exploration unleashed: The BIG framework for autonomous navigation

Teaming up tiny robot swimmers to transform medicine

Multiplexing entanglement in a quantum network

Mesoporous silicon: Semiconductor with new talents

Materials developed at Purdue University incorporated into new Microsoft Quantum qubit platform

Automatic cell analysis with the help of artificial intelligence

Quantum fractal patterns visualized

CSIC researchers discover how the brain builds sophisticated maps to navigate and remember the world

Supercomputing illuminates detailed nuclear structure

Material’s ‘incipient’ property could jumpstart fast, low-power electronics

Human chromosomes evolved at hyperspeed to give us better brains

AI generates playful, human-like games

Breakthrough in the development of a new low-cost computer

Investigating human interaction: When we are in sync

How the brain handles large text units

Study proposes a new theoretical framework for understanding complex higher-order networks

A blockchain-oriented covert communication technology with controlled security level based on addressing confusion ciphertext

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

Ocelot chip advances quantum computing

Amazon has announced Ocelot, its first quantum chip, developed by researchers at Amazon Web Services and Caltech.

Amazon claims that this chip marks a big step in building quantum computers that work well and don’t need too many parts. Ocelot uses superconducting circuits. This design helps Ocelot tackle quantum error correction, a way to fix mistakes in quantum computing caused by outside noise like heat or vibrations.

The chip brings three key improvements. First, it creates a scalable structure for bosonic error correction. This method relies on bosons, instead of simple two-state qubit systems, to more effectively protect quantum information from environmental noise and to do more efficient error correction.This beats older methods that need more parts to fix errors.

Second, Ocelot features a noise-biased gate, a tool that cuts down errors efficiently, making it easier to build bigger quantum computers. Third, it delivers high performance for superconducting qubits, tiny units that store quantum info. Ocelot’s qubits last nearly one second before bit-flip errors happen, and 20 microseconds before phase-flip errors occur.

A Caltech press release has more information. The researchers have described Ocelot and preliminary research results in a paper published in Nature. MIT Technology Review has covered the launch of Ocelot (unpaywalled copy).

Toward practical quantum computing

Quantum computers promise to solve tough problems way faster than regular computers. But today’s quantum machines struggle with noise, limiting them to about a thousand quantum gates before errors ruin everything.

Ocelot aims to bridge this gap with smarter error correction. By spreading one logical qubit’s info across several physical qubits, it shields data from noise and fixes mistakes.

Traditional error correction needs thousands of physical qubits per logical qubit, making big quantum computers hard to build. Ocelot changes that with cat qubits, named after a famous thought experiment. These qubits use oscillators to store info. Adding more energy, or photons, cuts bit-flip errors sharply, so fewer qubits are needed. Tests show Ocelot’s design works well, reducing errors as it scales up.

Amazon believes Ocelot could slash resource needs by 90% compared to older methods, paving the way for practical quantum computing that could transform society.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

Exploring animal consciousness with a new method

Researchers have come up with a fresh way to study animal consciousness. Their method might help us see how animals think and feel compared to humans.

The approach uses a marker method. This means looking at behaviors and body features tied to consciousness in people, then checking if animals have them too. Consciousness here means being aware of things like pain or sights. The researchers think this could shed light on big questions about what consciousness really is, maybe even helping us understand our own minds better.

The researchers argue that, if animals act like humans in ways tied to consciousness, it might show they’re conscious too.

The researchers have described the methods and results of this study in an essay published in Science.

More open-minded research is needed

Their essay builds on the “New York Declaration on Animal Consciousness.” That declaration, backed by over 500 scientists, claimed consciousness likely exists in all vertebrates, like birds and fish, and many invertebrates, like octopuses.

People have wondered about animal minds for a long time. Even today, experts don’t fully agree on how widespread consciousness is among animals. The researchers suggest focusing on specific parts of consciousness, like feeling pain or seeing things. They want scientists to look for clues, or markers, in different animals to see if those parts are there.

They also push for studying more than just pain, like maybe joy, and using methods that don’t harm animals. But they admit one marker alone isn’t enough proof. For example, talking is a sign of conscious thought in humans, but fancy computer programs can fake it without being conscious. Context matters a lot, they say. Still, they believe keeping an open mind and doing more research is key. One day, we might say with confidence that many animals are conscious. For now, the evidence isn’t perfect, so they urge everyone to keep exploring.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

New technology shapes flexible brain electrodes

Scientists found a new way to make microelectrode arrays, or MEAs. MEAs are tools that record brain activity and stimulate nerve tissues. Regular MEAs stay flat, so they struggle to fit the brain’s natural curves. Older methods to add 3D shapes take many steps, which complicates things and limits designs.

The new method, called μETF, draws from plastic thermoforming. Thermoforming molds plastic sheets into shapes by heating and pressing them.

A study published in npj Flexible Electronics describes the new method.

The scientists got the idea from watching how plastic coffee cup lids form. They thought this could work for tiny electrodes too.

With μETF, the scientists heat a thin, bendy sheet with tiny electrodes and press it onto a 3D-printed mold. They use liquid crystal polymer, or LCP, for the sheet. LCP is strong, safe for the body, and lasts a long time. This creates bumps and dips that bring electrodes closer to neurons, keeping the electrodes working well electrically. Unlike older ways, μETF makes complex shapes like wells, domes, and triangles in one go.

This technology might power brain-computer interfaces (BCIs)

The scientists tested μETF by making a 3D MEA for retinal stimulation in blind people. Tests and computer models showed these 3D electrodes work better than flat ones. They need 1.7 times less power and focus 2.2 times sharper. Closeness to neurons boosts efficiency and accuracy.

This technology could help beyond the retina. The scientists see it working for brain, spinal cord, ear, and nerve interfaces. They think this technology might power brain-computer interfaces, or BCIs. A 3D MEA in the motor cortex, which controls movement, could decode signals to move robotic arms or wheelchairs for paralyzed people.

The scientists plan to tweak the method for more medical uses. With simpler steps and better results, μETF pushes neural technology forward. It could improve treatments for brain and nerve problems.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

Tiny tech creates smallest shooting game

Researchers have used nanoparticles to build the tiniest shooting game ever. This game mixes real tiny objects with digital ones.

The researchers describe the methods and results of this study in a paper published in Japanese Journal of Applied Physics. Their goal is to make a computer system that connects virtual stuff with real nanomaterials smoothly.

The game combines digital technology with the real nanoworld using fast-moving electron beams. These beams create electric fields and light patterns on a screen. Researchers control these fields to move nanoparticles around in real time.

This is a fun shooting game like the arcade classics. Players use a joystick to steer a triangle-shaped spaceship on the screen. The spaceship comes from the electron beam’s pattern. Players try to hit enemy characters, which are actually little polystyrene balls. Polystyrene is a type of plastic made into nano-sized spheres here.

Electron-beam induced electro-force field display for dynamical biomanipulation system (Credit: The Japan Society of Applied Physics).

Players shoot at real nanoparticles to push them away

The researchers call this the “world’s smallest shooting game.” The system turns the digital ship into a light image and force field in real nano-space. Players shoot at nanoparticles to push them away. This proves digital data can work with real nano-objects instantly.

This technology isn’t just for fun. It could help scientists move and build tiny biological samples, like parts of cells. Nanotechnology, which is technology at the super-small scale, might benefit a lot. So could biomedical engineering, a field that mixes medicine and technology to solve health problems.

The researchers think this could change 3D printing and let engineers print tiny creations right as they design them. It could also guide harmful substances to attack virus cells in living things and destroy them. The ersearchers conclude that his tiny game opens big possibilities for science and medicine.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

AI tool helps scientists discover faster

Researchers at Monash University have built an artificial intelligence (AI) tool to help scientists work faster. They call it LLM4SD, short for Large Language Model for Scientific Discovery.

This tool acts like a scientist. It reads old research papers and studies data to make guesses about new findings. LLM4SD is available on GitHub with the description: “LLM4SD is an open-source initiative that aims to leverage large language models for scientific discovery.”

The researchers describe LLM4SD, as well as some preliminary results that the tool gas achieved, in a paper published in Nature Machine Intelligence.

This AI is a large language model (LLM) that handles basic research steps. It pulls useful facts from science writings and forms ideas from data. Unlike many tools, it explains its answers clearly. This helps scientists trust it.

LLM4SD is available on GitHub. “LLM4SD is an open-source initiative that aims to leverage large language models for scientific discovery,” reads the GitHub description. “We have now released the complete code.”

LLM4SD blends old knowledge with new machine learning

The researchers tested LLM4SD on 58 tasks about molecule traits. These tasks spanned four science areas: physiology, physical chemistry, biophysics, and quantum mechanics. The tool did well, beating other top tools by up to 48 percent in guessing quantum traits for designing materials.

In a Monash University press release, the researchers say LLM4SD it’s like ChatGPT but for science. It reads years of studies and lab results to answer questions. For example, it predicts if a drug can reach the brain or if a compound dissolves in water. It explains its steps simply, unlike older “black box” tools that hide their thinking.

The researchers say that LLM4SD makes predictions reliable and clear for all scientists. It blends old knowledge with new machine learning, combining knowledge and spotting patterns. For example, it quickly digs through old research and finds new clues.

The researchers claim that LLM4SD could simplify drug discovery and help scientists everywhere. It promises faster, easier, and sharper science breakthroughs.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

Tiny antennas hear low-frequency signals

Researchers have created a tiny antenna for low-frequency signals. Their work cuts antenna size by nearly 10,000 times compared to old designs. This could help solve big problems in making small antennas for tough jobs like underwater searches, underground scans, and space signal guides.

The researchers describe the methods and results of this study in a paper published in PhotoniX.

Low-frequency signals range from 30 to 300 kHz. They travel far, cut through barriers, and resist interference. Normal antennas struggle here. Smaller ones lose sensitivity fast. Older tiny antennas, like magnetoelectric ones, stay stuck at centimeter sizes because their frequency ties to their length.

Silica nanoparticles levitated by lasers and charged by electron beams

The researchers used silica nanoparticles, just 143 nanometers wide, held up by lasers in a vacuum. They improved the design in key ways. First, they zapped the particles with electron beams to add charge and make them very sensitive to electric fields. Second, the particles’ frequency comes from the laser’s power, not their size. So, a 100-nanometer antenna works from 30 kHz to 180 kHz. Third, they used a method called 2FSK modulation – shifting signal patterns – to read signals clearly, with almost no errors at 0.5 kbit/s in weak fields.

This antenna tunes its frequency by tweaking the laser. It senses signals better than 10 microvolts per centimeter per square root hertz. It tracks motion in 3D, catching signals from all directions. Tests showed it can send images with few mistakes, proving it works.

Still, it’s less sensitive than big antennas by a factor of 1,000 to 10,000. Its small size and flexibility shine in harsh spots like deep seas or tight spaces. The researchers plan to improve it and reach even lower frequencies with magnetic tricks or new materials. They also hope to fit it on chips for easy use.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

Robots learn about their bodies by watching themselves

Robots can now figure out their own body structure and movements by watching themselves with a camera. A new study from Columbia Engineering researchers shows this clearly. With this skill, robots plan their actions better and even handle damage to their parts.

The robots use raw video to understand their bodies, much like humans learning to dance by watching a mirror. The goal is simple. The researchers want robots to adapt to damage and learn new tasks without humans always stepping in.

The study is published in Nature Machine Intelligence.

Usually, robots start learning in simulations, which are virtual spaces. After practicing there, they move to the real world. Good simulators make this switch easier. However, building these simulators takes a lot of effort from skilled engineers. In this study, the researchers found a shortcut. They taught a robot to make its own simulator just by watching its motions on camera. This saves time and lets the simulator grow with the robot.

The researchers used a single 2D camera to help the robot model its 3D shape. Deep neural networks powered this process. The neural networks turned 2D video into 3D motion clues, helping the robot adapt. The system also spotted changes, like a bent arm, and adjusted the robot’s movements to keep it going.

Robots could imagine their future actions

The researchers imagine a robot vacuum with a bent arm after hitting a chair. Instead of stopping, it watches itself, tweaks its moves, and keeps cleaning. In a car factory, a robot arm might get misaligned. Rather than pausing work, it adjusts itself and resumes welding, saving time and money.

Robots must handle critical tasks, like in factories or hospitals. Humans can’t always fix them. Self-modeling helps robots care for themselves and stay useful. Now, with one camera and a short video, robots gain what researchers call “Kinematic Self-Awareness.” Robots could imagine their future actions, unlocking endless possibilities.

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter. 

Mindplex is a magazine dedicated to the rapidly unfolding future, a platform to enable our community to share and discuss futurist content, and a showcase for AI tools to improve the media experience. Join us now and help create the future of digital media!

Quick links

About us

Contact

© 2022 Mindplex all rights reserved

Download our app