Weekly Piece of Future #87
From Bioengineered Immune Cells to Nvidia’s NVLM Open-Source Model and Perovskite Solar Cells with 24.13% Efficiency
Hey there, fellow future-addicts!
Welcome to this week's edition of Rushing Robotics, where we delve into the latest breakthroughs and transformative innovations shaping the future of artificial intelligence and biotechnology!
🤯 Mind-Blowing
Nvidia has launched NVLM 1.0, an open-source AI model family with a 72-billion-parameter model, challenging industry titans like OpenAI and Google. Meanwhile, Stanford researchers have unveiled new methods for producing neurons in aging brains, offering hope for future treatments of degenerative diseases. Engineers have also pioneered a revolutionary approach to robotic control using fluid circuits instead of electricity. And in robotics, AI-generated images are training machines more efficiently than ever before.
🔊 Industry Insights & Updates
This week brings big news, with Google DeepMind scientists and AI pioneers like Geoffrey Hinton receiving the Nobel Prize for their transformative work in protein structure prediction and machine learning. Tesla is set to unveil its autonomous robotaxi, signaling major strides in self-driving tech, and China is making significant headway in silicon photonics, paving the way for advanced chip design in light-based data transmission.
🧬 BioTech
Biotech is buzzing with innovations: researchers have found a way to accelerate scar-free healing using proteins from a parasitic worm, while a new paper-based sensor developed by UCLA can diagnose heart attacks in just 15 minutes. In a groundbreaking first, CRISPR-modified immune cells have brought autoimmune disorders into remission in patients in China.
💡 Products/Tools of the Week
Check out BaseAI, a powerful framework that simplifies AI-powered product development, CostGPT AI, a tool that quickly generates project plans and cost estimates, and Trupeer.ai, which transforms screen recordings into professional training and demo videos. Plus, Papers helps streamline academic research with AI-powered recommendations and seamless citation management.
🎥 Video Section
This week’s video section highlights some of the most exciting advancements in robotics and AI. From a robot built to play first-person shooter games to humanoid robots engaging in conversation, these videos showcase the growing capabilities of intelligent machines. You’ll also see a quadrupedal robot demonstrating impressive agility by climbing a ladder. Dive in to witness the future of robotics and AI in action.
Join us as we explore these mind-blowing updates and much more, revealing the incredible potential of AI to reshape industries and improve lives. Stay hungry, stay futurish!
🤯 Mind-Blowing
Nvidia has launched a powerful open-source artificial intelligence model that positions itself as a competitor to proprietary systems from major players like OpenAI and Google. The new NVLM 1.0 family of large multimodal language models, spearheaded by the 72 billion parameter NVLM-D-72B, showcases outstanding performance across both vision and language tasks, while also improving text-only capabilities. By making the model weights publicly available and committing to release the training code, Nvidia is breaking away from the common practice of keeping advanced AI systems under wraps. Notably, the 72B model performs comparably to OpenAI’s Llama 3.1 405B in math and coding evaluations and also incorporates vision functionality.
A study from Stanford Medicine reveals promising insights into generating new neurons in aging brains. The researchers indicate that it may be possible to develop pharmaceutical or genetic therapies that can stimulate the production of new neurons in older or injured brains. Typically, neurons in the human brain endure for a lifetime, maintaining vital information through intricate structural connections within their synapses. The loss of these neurons would mean a loss of critical information, leading to forgetfulness. While some new neurons are still produced in the adult brain by neural stem cells, their ability to do so declines with age. This decline can have severe neurological effects, impacting memory and contributing to degenerative brain diseases. The research also proposes exciting future directions for addressing the inactivity of old neural stem cells by targeting newly discovered pathways that could reactivate them.
Engineers have devised a groundbreaking method for transmitting complex instructions to robots without relying on electricity, potentially freeing up more system processing space for advanced functions. Researchers at King’s College London (KCL) created a novel micro-circuit that communicates through changes in fluid pressure, inspired by the human body. This innovative approach allows robots to execute tasks without conventional electrical systems by using fluid to transmit commands within the circuit. According to the team, this "world first" could lead to a new generation of robots, where their bodies operate independently of a control center, making room for more sophisticated AI-driven software.
Researchers have introduced a groundbreaking approach to training robots using AI-generated images. The system, known as Genima, fine-tunes the Stable Diffusion model to generate training data that guides robotic movements in both simulations and the real world. Developed by the Robot Learning Lab in London, led by Stephen James, this method could simplify training for a wide range of robots, from mechanical arms to humanoid robots and autonomous vehicles. Genima acts as a behavior-cloning agent, producing joint-action drawings on RGB images that are mapped into sequences of joint positions. Tested on 25 RLBench and 9 real-world tasks, the system has shown superior performance over existing visuomotor methods, particularly in handling scene changes and adapting to new objects.
Researchers in South Korea have developed a groundbreaking passivation process for formamidinium lead iodide (FAPbI3) perovskite films, achieving a certified power conversion efficiency of 24.13% in solar cells. Led by Professor Hobeom Kim from the Gwangju Institute of Science and Technology (GIST), the research team implemented a novel defect passivation strategy that significantly reduces defects while enhancing both the efficiency and stability of perovskite solar cells. Their findings indicate that integrating hexagonal polytype (6H) perovskite with cubic polytype (3C) FAPbI3 resulted in a substantial increase in power conversion efficiency compared to traditional methods.
🔊 Industry Insights & Updates
Three scientists, including two from Google DeepMind, have been awarded the 2024 Nobel Prize in Chemistry for groundbreaking achievements in protein structure prediction and design. Their work has revolutionized fields such as medicine and nanotechnology, highlighting the transformative potential of AI and computational techniques in modern science. Demis Hassabis, founder of Google DeepMind, and John Jumper, a lead developer of the AlphaFold model, share the prestigious prize for their pivotal contributions.
On Tuesday, two pioneers in artificial intelligence, John Hopfield and Geoffrey Hinton, were awarded the Nobel Prize in Physics for their groundbreaking work in machine learning. Their contributions have transformed the way we live and work, though one of the winners also warned of potential dangers posed by this technology. Hinton, often called the “Godfather of AI,” holds both Canadian and British citizenship and is based at the University of Toronto. John Hopfield, an American scientist, works at Princeton University. Their research has formed the backbone of the machine learning technologies that drive today’s AI innovations.
Tesla is set to unveil its self-driving robotaxi, an autonomous vehicle designed to solidify the company's leadership in AI and robotics. The reveal will take place at Warner Bros. Studios in Burbank, California, where Tesla has reportedly been gathering updated mapping data ahead of the event. With competition from rivals like Waymo and Cruise, Tesla faces the challenge of proving its readiness to launch a driverless vehicle. Elon Musk is also expected to present his vision for the Tesla Network, allowing Tesla owners to add their autonomous cars to a robotaxi fleet when not in use.
Wuhan-based JFS Laboratory, a state-funded semiconductor research facility, has announced a significant breakthrough in silicon photonics technology. Silicon photonics, which uses light for data transmission, is seen as a way to overcome the physical limits of electrical signals. This development could help China address current chip design challenges and advance technological independence despite US sanctions. According to a recent blog post, the lab successfully integrated a laser light source with a silicon-based chip for the first time in China, marking a major advancement in the nation's optoelectronics technology.
🧬 BioTech
Scientists have made a fascinating breakthrough that could enable scar-free healing. A protein sourced from Heligmosomoides polygyrus, an intestinal parasitic roundworm, has been found to accelerate wound healing while minimizing scarring in mice. This discovery offers promising new approaches for treating injuries without the unwanted side effects of scar tissue formation. While human skin is adept at healing small cuts, deeper injuries to the dermis often lead to scarring, which can cause lasting changes in appearance and limit movement. Researchers have long sought ways to improve natural healing without the formation of scars, and this new finding could bring them closer to that goal.
Heart attacks remain one of the leading causes of death worldwide, and early diagnosis is critical to saving lives. However, standard lab tests are often slow and costly. In response, researchers from the University of California, Los Angeles (UCLA) have developed a low-cost, paper-based sensor that can detect heart attack biomarkers. This new device delivers results in just 15 minutes. According to the researchers, the sensor offers exceptional sensitivity, surpassing that of current point-of-care devices. It meets clinical standards for detecting small concentrations of troponin protein in the blood, potentially enabling the early diagnosis of acute myocardial infarction (AMI), or heart attacks.
China has achieved a world-first in treating autoimmune disorders, with three individuals experiencing remission after receiving bioengineered and CRISPR-modified immune cells. The patients, two men and one woman, were the first to undergo treatment using donor-derived immune cells, rather than cells from their own bodies. One patient, named Gong, who suffered from systemic sclerosis—a disease affecting connective tissue, leading to skin stiffening and organ damage—reported improvement just three days after receiving the therapy.
💡Products/tools of the week
BaseAI is a web-based AI framework that simplifies the process of building, configuring, and deploying AI-powered products using Large Language Models (LLMs). It allows developers to create AI agent pipes, which are customizable workflows combining diverse AI tools and memory modules. BaseAI emphasizes flexibility and composability, supporting multiple LLMs and offering streamlined deployment via Langbase, a serverless AI cloud. Developers appreciate BaseAI's simplicity, scalability, and modular approach, which allows them to build complex AI systems without needing in-depth AI expertise or managing intricate infrastructure.
CostGPT AI is an AI-driven tool that simplifies software project planning by generating detailed plans, cost estimates, and timelines from a brief project description. Tailored for project managers, developers, and business owners, it quickly produces comprehensive project roadmaps, including feature breakdowns, sitemaps, user stories, and key milestones. This tool is ideal for those looking to save time, enhance accuracy in project estimation, and gain insights that support efficient project management and resource allocation.
Trupeer.ai is an advanced AI platform that effortlessly converts screen recordings into professional product videos, demo presentations, training materials, and documentation. It automates the creation of scripts, voiceovers, and visuals, dramatically reducing the effort involved in content production. Designed for product marketers, support teams, technical writers, and learning and development professionals, Trupeer.ai enables the rapid creation of high-quality, customizable content, eliminating the need for extensive editing experience or writing expertise.
Papers is a sophisticated reference management platform designed for students, academics, and professionals who need to efficiently manage academic resources. It simplifies the processes of sourcing, organizing, and sharing research by allowing users to import data from desktops, databases, and journals. With intuitive metadata matching and management tools, Papers offers an excellent experience for organizing collections. In addition to supporting existing research management, it helps users discover new studies, providing AI-driven article recommendations and access to a large, searchable database. Furthermore, it assesses research impact through linked citations, Altmetrics, and Retraction Watch, while offering annotation and note-taking features for detailed analysis.
