Category: AI

Mother Nature is widely regarded as the most powerful generative force, having designed the vast and intricate variety of life on Earth using just four genetic letters—A, T, C, and G. But can generative AI build upon her work?

A groundbreaking new algorithm called Evo 2 is pushing the limits of what AI can achieve in the realm of biology. Trained on an immense dataset of roughly 128,000 genomes—equating to 9.3 trillion DNA letter pairs from all domains of life—Evo 2 is now the largest generative AI model ever created for biological research. Developed by scientists at the Arc Institute, Stanford University, and Nvidia, Evo 2 is capable of writing entire chromosomes and small genomes from scratch.

Figure AI Inc. has recently showcased the capabilities of its Helix visual-language-action (VLA) model in a domestic setting, demonstrating the impressive potential of humanoid robots in everyday tasks. In a video released by the company, the robots are tasked with putting away groceries, a seemingly simple chore that becomes a showcase for advanced robotic collaboration and AI-driven coordination.

The demonstration begins with a single prompt from a human, and from there, the robots visually assess the scene, identifying each item and then cooperatively placing them in their designated locations in the kitchen. This marks an important milestone for Figure AI’s robotic systems and reveals a couple of notable takeaways.

This infographic explores how artificial intelligence (AI) is being utilized in various occupations across the United States. It examines the proportion of AI-related conversations within each job category and compares these figures with each group’s share of the overall U.S. workforce.

The data comes from Anthropic’s analysis of millions of conversations on Claude, an AI-powered chat assistant developed by Anthropic. The findings are detailed in their research paper, Which Economic Tasks Are Performed with AI? Evidence from Millions of Claude Conversations.

Artificial intelligence (AI) has been making headlines recently for its ability to write essays, generate art, and even pass medical exams. But despite these impressive feats, most AI systems still need significant human guidance to operate effectively. Much like a student who requires constant instruction, today’s AI depends heavily on meticulously labeled data and rigid rules to learn. Now, researchers at the University of Technology Sydney have developed an innovative method that brings AI closer to the way humans and animals learn naturally. This breakthrough approach could enable AI to learn independently by identifying patterns within data—without the need for explicit instructions.

“As in nature, animals learn by observing, exploring, and interacting with their environment, without detailed guidance,” explains Distinguished Professor CT Lin from the University of Technology Sydney. “The next wave of AI, known as ‘unsupervised learning,’ aims to emulate this more organic process.”

A UK-based company, SECQAI, has made a groundbreaking announcement with the launch of the world’s first Quantum Large Language Model (QLLM), a move that promises to shape the future of artificial intelligence (AI). By integrating quantum computing into traditional AI frameworks, SECQAI aims to significantly enhance computational efficiency, problem-solving capabilities, and linguistic understanding in large language models.

The QLLM was developed by combining quantum computing with classic AI models, using an in-house quantum simulator with gradient-based learning and a quantum attention mechanism. This marks a significant step in the convergence of AI and quantum technology, as SECQAI believes the launch will usher in a new era of quantum-powered AI solutions.

Researchers at the Institute of Molecular Biology and Biotechnology (IMBB) of FORTH have unveiled a groundbreaking artificial neural network (ANN) model that draws inspiration from biological dendrites. This novel approach promises to revolutionize image recognition systems by drastically reducing the number of parameters needed, making AI more compact, energy-efficient, and accessible.

Artificial intelligence is rapidly transforming industries by offering advanced solutions to complex challenges, yet most AI systems today require vast amounts of computational power. Current models often consist of millions to billions of parameters, leading to high energy consumption and large-scale infrastructure needs. These inefficiencies limit the potential for widespread adoption, especially in resource-constrained environments.

A team of researchers from South Korea has developed an innovative fire detection technology designed to drastically reduce false alarms—those triggered by non-fire-related particles like dust or steam—which have been a persistent issue in fire safety systems. This groundbreaking technology is on the verge of commercialization and is expected to significantly cut the social costs caused by these “unwanted alarms.”

The Electronics and Telecommunications Research Institute (ETRI) recently unveiled an AI-powered sensor capable of distinguishing between smoke from an actual fire and non-fire aerosol particles. The system works by analyzing the scattering patterns of light across particles of varying wavelengths, enabling it to differentiate between harmless particles and those associated with a real fire.

Our modern world is powered by computer chips, from the ones in cars and smartphones to those that help track animals and optimize various industries. As technology continues to advance, there is a constant push to develop faster, more efficient, and innovative chips. To achieve this, some researchers are turning to artificial intelligence (AI) to assist in the chip design process, sometimes taking a step back from human control.

A team of scientists recently shared their process of allowing AI technology to design and test more efficient computer chips. Led by Kaushik Sengupta, an electrical engineer at Princeton University, the research explores the potential of using AI to enhance chip development. Sengupta, who was recently awarded an IEEE fellowship for his work on wireless chip technology, emphasizes that the goal is not to replace human engineers, but rather to augment their productivity. By publishing their findings in the open-access, peer-reviewed journal Nature Communications, Sengupta’s team is making their AI-driven research available to the broader scientific community.

Radiologists are increasingly relying on AI-based computer vision models to assist with the time-consuming task of interpreting medical scans. However, these AI models require vast amounts of accurately labeled data to function effectively, meaning radiologists must still invest significant time annotating medical images. To address this challenge, an international team led by Johns Hopkins Bloomberg Distinguished Professor Alan Yuille has developed AbdomenAtlas, the largest abdominal CT dataset to date. With over 45,000 3D CT scans and 142 annotated anatomical structures from 145 hospitals worldwide, AbdomenAtlas is more than 36 times larger than its nearest competitor, TotalSegmentator V2. This remarkable dataset and its findings were published in Medical Image Analysis.

Historically, abdominal organ datasets were created through the labor-intensive process of having radiologists manually label each individual organ in CT scans. This process required thousands of hours of expert labor. “Annotating 45,000 CT scans with 6 million anatomical shapes would require an expert radiologist to have started working around 420 BCE—the era of Hippocrates—to complete the task by 2025,” explains lead author Zongwei Zhou, an assistant research scientist at Johns Hopkins University.

Synthesizing new proteins, which are fundamental to all biological life, is an area of immense scientific promise. A groundbreaking development from researchers in the U.S. has taken a major step forward in this field with the use of an advanced AI model called EvolutionaryScale Model 3 (ESM3). This model has been used to create a new protein, called esmGFP (green fluorescent protein), which shares just 58 percent of its material with its closest natural relative, tagRFP.

The research team estimates that this breakthrough represents the equivalent of processing 500 million years of evolution via AI, opening new doors to creating custom-made proteins designed for specific applications, or enhancing the functions of existing proteins. According to the researchers, led by Thomas Hayes, founder of EvolutionaryScale in New York, “More than three billion years of evolution have produced an image of biology encoded into the space of natural proteins.” Their study demonstrates how language models trained on evolutionary data can generate functional proteins that are significantly distant from any known natural proteins.

Researchers from Zhejiang University and HKUST (Guangzhou) have developed an advanced AI model, ProtET, that harnesses the power of multi-modal learning to enable controllable protein editing through simple text-based instructions. This breakthrough, detailed in Health Data Science, bridges the gap between biological language and the manipulation of protein sequences, advancing functional protein design across various domains, such as enzyme activity, stability, and antibody binding.

Proteins are vital to all biological processes, and their precise modification holds tremendous potential in areas like medical therapies, synthetic biology, and biotechnology. Traditional methods of protein editing typically involve time-consuming laboratory experiments and single-task optimization models. However, ProtET introduces a transformative approach using a transformer-structured encoder and a hierarchical training paradigm. The model aligns protein sequences with natural language descriptions through contrastive learning, allowing researchers to modify proteins intuitively using text-based instructions.

DeepSeek, a Chinese AI research lab, has recently shocked the tech world by unveiling its groundbreaking AI model, DeepSeek-R1, which achieves performance levels comparable to the best chatbots in the world—but at a fraction of the cost. This one-year-old company has thrown into question the prevailing notion that developing cutting-edge AI requires ever-increasing financial and energy investments, positioning itself as a serious competitor to established players like OpenAI, Google, and Meta.

The launch of DeepSeek’s open-sourced AI model has caused a ripple effect across the global tech market, with stocks of US-based technology giants tumbling in response. DeepSeek’s sudden rise has raised alarms about the US’s position as the dominant force in AI development, as well as the valuations of major tech companies.