Weekly Piece of Future #105
From Lab-Grown Muscles to Advanced Prosthetics and Hybrid Quantum Supercomputers
Hey there, fellow future-addicts!
Welcome to this week's edition of Rushing Robotics! This week, we explore biohybrid robotics, AI-powered breakthroughs in mathematics, quantum computing milestones, and next-gen prosthetics. Whether you're fascinated by futuristic technology, industry trends, or life-changing biotech, this issue has something for you!
🤯 Mind-Blowing
Japanese researchers have developed a robotic hand powered by real muscle tissue, marking a major step toward more lifelike prosthetics. Meanwhile, DeepMind’s AI is solving geometry problems better than Olympiad gold medalists, and quantum supercomputing is entering a new era with revolutionary hybrid systems. These breakthroughs are redefining what AI and robotics can achieve.
🔊 Industry Insights & Updates
The AI race is heating up! Mistral’s Le Chat is now the fastest AI assistant, humanoid robots are scaling up for real-world deployment, and China’s automakers are integrating advanced AI models into next-gen smart vehicles. Plus, an Austin-based robotics company just secured $350 million to bring its humanoid robots to the workforce. The future of AI-powered industries is unfolding faster than ever.
🧬 BioTech
In healthcare, researchers have introduced PAC-MANN, a new blood test for early pancreatic cancer detection, while the FDA has given the green light for gene-edited pig kidneys to be transplanted into humans. Plus, neuroscientists are uncovering the brain’s hidden decision-making mechanisms—insights that could revolutionize AI-driven technologies like autonomous systems and virtual assistants.
💡 Products/Tools of the Week
Want to explore the latest AI tools? This week, we highlight Venice AI, a privacy-first AI assistant, Skyvern, an automation tool for web-based tasks, and Storyblocker, a real-time 3D previsualization tool for filmmakers. Whether you’re a developer, creator, or business leader, these innovations are worth checking out.
🎥 Video Section
This week’s must-watch videos feature cutting-edge AI research, humanoid robots in action, and the latest breakthroughs from Meta, OpenAI, and UBTECH Robotics.
We’re witnessing the dawn of new technologies that will shape the future. Keep your eye on these developments—they’re just the beginning! Stay hungry, stay futurish!
🤯 Mind-Blowing
Japanese researchers have made a major breakthrough in biohybrid robotics by developing a robotic hand powered by lab-grown muscle tissue. This collaborative project, led by scientists from the University of Tokyo and Waseda University, employs multiple muscle tissue actuators (MuMuTAs)—bundles of engineered muscle strands capable of contracting with enough force to move the fingers. This innovation marks a significant step forward in biohybrid prosthetics and robotics, with potential applications in medicine, rehabilitation, and drug testing. The hand, built with a 3D-printed plastic base and muscle-based tendons, can perform gripping and scissor-like motions. Unlike previous biohybrid models, which were limited to single-joint movements, this new design enables multiple joints to operate independently or in coordination, greatly enhancing its versatility.
Google DeepMind, the AI research division of Google, has developed an AI system that appears to outperform the average gold medalist in solving geometry problems in international mathematics competitions. Named AlphaGeometry2, this system is an enhanced version of AlphaGeometry, which DeepMind originally introduced last January. According to a newly published study, AlphaGeometry2 can solve 84% of all geometry problems from the past 25 years in the International Mathematical Olympiad (IMO), a prestigious contest for high school students. Mathematical theorem proving—logically demonstrating why a theorem (such as the Pythagorean theorem) holds true—requires reasoning and selecting the correct problem-solving steps. If DeepMind’s claims hold, these capabilities could become a fundamental part of future general-purpose AI models.
A research team at the University of Science and Technology of China (USTC) has developed a high-dexterity biomimetic prosthetic hand, capable of executing complex motor functions, such as combing hair, interacting with smartphones, and performing sign language gestures. This lightweight prosthetic device, which closely mimics human hand mechanics, marks a significant advancement in prosthetic technology and humanoid robotics, with potential benefits for millions of amputees. The human hand possesses 23 degrees of freedom (DOFs) and accounts for 54% of total body functionality, despite comprising only one-150th of body weight. Conventional prosthetic hands, predominantly motor-driven, are constrained by weight and functional limitations, with most exceeding 0.4 kg and offering fewer than 10 DOFs. These shortcomings often lead to high user abandonment rates. To address these challenges, USTC researchers employed shape-memory alloys (SMAs)—materials that revert to a predefined shape when heated—to serve as artificial muscles. These SMAs, combined with a tendon-based transmission system, enhance force output while minimizing resistance.
Additionally, the team integrated 23 sensor units across the fingers and wrist to enable precise movement control and incorporated 38 SMA actuators with an advanced cooling module. The resulting prosthetic hand weighs just 0.37 kg, making it lighter than a human hand while achieving 19 active DOFs, significantly surpassing traditional designs.
Physicists have successfully constructed a new type of digital-analogue quantum simulator in Google’s laboratory, designed to study physical processes with an unprecedented level of precision and flexibility. Two physicists from PSI’s Center for Scientific Computing, Theory, and Data played a critical role in this achievement, contributing significantly to the research and development of this advanced simulator. The quantum simulator consists of 69 superconducting qubits and provides scientists with the ability to analyze quantum processes that remain beyond the computational reach of even the most powerful supercomputers on Earth. By leveraging the unique properties of quantum mechanics, this simulator opens new avenues for studying complex quantum phenomena that were previously inaccessible.
Engineers in Japan have successfully activated Reimei, the world’s first hybrid quantum supercomputer. This 20-qubit quantum machine has been integrated into Fugaku, currently the sixth-fastest supercomputer in the world.
The hybrid system is designed to tackle problems that classical supercomputers struggle to process efficiently. Located at the Riken scientific institute in Saitama, near Tokyo, Reimei will primarily support physics and chemistry research. Representatives from Quantinuum, the makers of Reimei, and Riken emphasized that this integration marks a major milestone in high-performance computing. Quantum computers have the potential to solve complex calculations in seconds or minutes, tasks that would take classical computers millions of years. However, since fully scalable quantum computers are still in development, scientists believe that hybrid systems like Reimei-Fugaku can act as an intermediate solution, offering practical quantum advantages today.
🔊 Industry Insights & Updates
Cerebras Systems is powering Mistral’s latest AI assistant, Le Chat, which is being hailed as the "world’s fastest AI assistant." As the competition for delivering high-performance, cost-effective AI intensifies—especially following China’s DeepSeek disruption in January—France’s Mistral is positioning itself as a key contender in the open-source AI space. Silicon Valley-based Cerebras, backed by UAE tech giant G42, has reportedly helped Mistral achieve a new AI speed record. The company’s newly launched Le Chat app can generate responses at an astonishing rate of 1,000 words per second, according to Mistral. Arthur Mensch, CEO and co-founder of Mistral, highlighted that this launch reinforces Europe’s growing significance in the AI sector.
An Austin-based company is gearing up to expand the production of humanoid robots designed to collaborate with humans. With new funding, Apptronik is focused on scaling its operations, driving innovation, and growing its expert team to meet the rising demand for AI-powered humanoid robots. Additionally, the company is accelerating the deployment of Apollo, its cutting-edge humanoid robot. Apptronik recently closed a $350 million Series A funding round, co-led by B Capital and Capital Factory, with Google also participating. “We’re developing the world’s most advanced and capable humanoid robots, designed to work alongside humans in meaningful and transformative ways,” said Jeff Cardenas, CEO and co-founder of Apptronik.
With DeepSeek’s rapid rise to over 20 million daily active users, Chinese automakers are now competing to integrate its AI model into their next-generation smart vehicles. Xpeng CEO He Xiaopeng has underscored the significance of AI in the automotive industry, stating, “In the next decade, AI will drive changes in the automobile industry that far exceed electrification.” Li Xiang, chairman of Li Auto, reinforced this perspective, asserting that electrification marks only the initial phase of transformation, while artificial intelligence represents the crucial second phase and the future of automotive design.
Heat2Power envisions its technology as a game-changer for industrial users, offering ultra-low-cost power in an environmentally sustainable way. The company asserts that its highly efficient TPV devices can integrate with high-temperature thermal energy storage (TES) systems, enabling cost-effective and reliable on-demand energy for renewable sources like wind and solar. Additionally, H2P’s technology provides an economical solution for recycling wasted energy commonly lost in industrial power plants. The company claims that its TPV technology outperforms all existing heat-to-electricity conversion devices in efficiency, functioning effectively at scales ranging from kilowatts to megawatts. Unlike other technologies that lose efficiency at lower temperatures, H2P’s solution remains stable across a broad temperature range.
🧬 BioTech
Researchers at Oregon Health & Science University (OHSU) have introduced PAC-MANN, a novel blood test designed for the early detection of pancreatic cancer. Given that pancreatic cancer has one of the highest mortality rates due to late-stage diagnosis, this test represents a significant advancement in the field. PAC-MANN, which stands for Protease Activity-Based Assay Using a Magnetic Nanosensor, specifically targets pancreatic ductal adenocarcinoma (PDAC), the most prevalent and aggressive form of pancreatic cancer. By enabling clinicians to detect the disease at an earlier stage, PAC-MANN may significantly improve treatment efficacy and patient survival outcomes.
Two U.S. biotechnology companies, United Therapeutics and eGenesis, have received FDA clearance to begin clinical trials testing gene-edited pig kidneys for human transplantation. This approval comes as the U.S. faces a severe organ shortage, with over 100,000 people awaiting transplants—more than 90,000 of whom need kidneys. United Therapeutics announced that the initial trial will involve six patients with end-stage renal disease, with the potential to expand to 50 participants based on early results. The first transplant is expected to take place by mid-2025.
Rapid decision-making in the human brain requires the integration of vast amounts of sensory information. A fundamental question in neuroscience is how the brain processes conflicting and redundant sensory cues to optimize decision outcomes. A recent study by Princeton neuroscientists provides novel insights into this process, with implications for both cognitive science and artificial intelligence. Their findings may contribute to advancements in AI-driven technologies such as autonomous vehicles and virtual assistants. The researchers proposed the ‘latent circuit’ model, which hypothesizes that decision-making is governed by a select subset of neurons, termed ‘ringleader’ neurons, rather than being distributed across the entire neural network. By analyzing recorded neural activity, they identified a key computational principle: when motion served as the primary cue, motion-processing neurons actively suppressed color-processing neurons. Conversely, when color was the dominant sensory input, color-related neurons inhibited those responsible for motion processing. This ‘low-dimensional’ framework offers a paradigm shift in how neural computations are conceptualized. To empirically test their model, Langdon and Engel employed a decision-making paradigm commonly utilized in human and animal research.
💡Products/tools of the week
Venice AI puts privacy first, giving you a decentralized AI platform for generating text, creating images, and assisting with code—all without storing your data. Unlike traditional AI tools, it offers a secure and unrestricted experience, making it perfect for users who value privacy and free expression. Whether you’re a developer, researcher, content creator, or just someone who wants full control over their AI interactions, Venice AI delivers a customizable and censorship-free experience.
Skyvern is an AI-powered browser automation tool that leverages computer vision and large language models to interact with web pages and execute tasks such as data extraction, form filling, and managing complex workflows. It is specifically designed for businesses aiming to automate repetitive web-based processes with greater efficiency. Unlike conventional automation tools, Skyvern dynamically adapts to evolving website structures and functions on previously unseen sites without requiring custom coding. This makes it a highly versatile and powerful solution for organizations looking to optimize operations and scale efficiently while minimizing costs.
Storyblocker is an innovative previsualization tool designed for filmmakers and content creators, integrating Augmented Reality, Real-Time 3D, and AI to optimize the preproduction process. With features such as 3D character manipulation, set building, and AR-powered scene projection, users can storyboard, block scenes, and animate directly from their smartphones. By providing a detailed and realistic visualization of projects before production, Storyblocker helps filmmakers save time, reduce costs, enhance creative control, and improve collaboration.
A0.dev is an AI-powered tool designed to generate user interface code based on either text prompts or design inputs. It likely focuses on creating React components utilizing libraries such as Tailwind CSS, much like Vercel’s v0.dev. This tool serves as a valuable resource for developers and designers by enabling rapid prototyping of UI elements, optimizing the development workflow, and bridging the gap between design and code implementation. By automating UI code generation, A0.dev not only saves time but also promotes consistency in web application development.
AI at Meta: Meta PARTNR: Unlocking Human-Robot Collaboration
Humanoid: Introducing HMND 01 | Our First Next-Generation Humanoid Robot
AI News: Phantom Humanoid Robot MK1 Is The FIRST AI Android Training For War (OPENAI DEEP RESEARCH)
Kevin Wood | Robotics & AI: ELEGNT: Apple Robot Lamp Companion with Expressive Humanlike Motions
UBTECH Robotics: UBTECH Humanoid Robot Walker S: Workstation Assistant in EV Production Line