Weekly Piece of Future #39
From Brain-Computer Interfaces to LLMs with Infinite Context Length and Data Storage for 10,000 Years
Hey there, fellow future-addicts!
Welcome to this week's edition of Rushing Robotics, your source for the most incredible developments in the world of artificial intelligence and technology. In this issue, we're excited to bring you stories that highlight the awe-inspiring possibilities of human-machine collaboration and cutting-edge materials. From brain-computer interfaces that empower ALS patients to control their surroundings with their thoughts, to plant-based materials driving the creation of tiny, intelligent robots for medical procedures, the future is truly here.
We'll also delve into the world of advanced language models with UC Berkeley's MemGPT, a processor that goes beyond fixed context, and Microsoft's Project Silica, which promises data storage that can last for millennia. And in the field of healthcare, we explore the development of a cancer survival calculator powered by AI, offering personalized prognosis estimates for cancer patients.
But the excitement doesn't stop there. Our "Industry Insights & Updates" section features NVIDIA's groundbreaking approach to teaching robots complex tasks through AI, Toyota's potential game-changing solid-state battery, and the buzz around the Cybertruck-inspired electric scooter, P1.
In "BioTech," we share a revolutionary medical procedure for kidney stone removal, a superconducting camera that's set to transform scientific research, and a promising drug in the fight against neuroendocrine tumors.
And, of course, we have a selection of remarkable products and tools to enhance your daily life, from Scribot's innovative note-taking capabilities to Echobase's advanced AI model training platform, and the ultimate all-in-one coding assistant, CodeCompanion.AI. Plus, learn how Arc is reshaping remote software developer hiring with AI technology.
Don't forget to check out our video section for the latest developments in robotics, from Boston Dynamics' Multipick for Stretch to KIMLAB's vacuum robot with an arm, Verity's drone technology at IKEA, and Agility Robotics' humanoid robot's incredible resilience.
In the ever-evolving realm of AI and Robotics, we're committed to keeping you informed. Until next time, keep exploring, learning, and creating! Stay hungry, stay futurish!
🤯 Mind-Blowing
A remarkable brain-computer interface has successfully reestablished control over home devices for ALS patients. This innovative BCI enables individuals with ALS to effortlessly control household appliances using only their thoughts. Notably, the brain-implanted device demonstrated consistent control over communication boards, lights, and televisions for an impressive three months without requiring recalibration.
Unlocking new possibilities, plant-based materials are the driving force behind the creation of small, intelligent robots by a team of University of Waterloo researchers. These cutting-edge materials are poised to be the cornerstone for the next wave of soft medical microrobots. With their potential to conduct minimally invasive medical procedures such as biopsies and cell/tissue transport, these tiny robots mark a significant step forward in the world of healthcare.
UC Berkeley's MemGPT, a fascinating Large Language Model (LLM), that breaks the boundaries of context. This unique LLM processor, equipped with a tiered memory system and self-management functions, extends its capabilities beyond the standard fixed context. The primary context remains the fixed-length LLM input, but MemGPT excels in parsing LLM text outputs, offering the choice to either relinquish control or initiate a function call. These functions serve the purpose of transferring data between the main and external contexts.
Microsoft's Project Silica is making data storage on glass plates a reality, with an astonishing lifespan of 10,000 years. A single, small glass disk can accommodate a substantial 3,500 films or an incredible 1.75 million songs. This marks a significant advancement in data storage, offering superior durability and simplicity compared to conventional hard disks.
Revolutionizing cancer care, researchers are working on a cancer survival calculator powered by artificial intelligence. This sophisticated machine learning tool is designed to offer personalized prognosis estimates for patients dealing with breast, thyroid, and pancreatic cancers, providing them with valuable information to navigate their treatment paths.
🔊 Industry Insights & Updates
An exciting development in AI and robotics: NVIDIA Research has introduced a novel method for teaching robots complex tasks. Through Large Language Models (LLMs), an AI agent generates reward algorithms, empowering robots to master challenging activities autonomously.
Toyota announces a significant battery breakthrough that could potentially revolutionize electric cars. The Japanese company is confident in its ability to develop a solid-state battery with an impressive range of 745 miles and the capability to charge in just 10 minutes. It is likely to go into mass production by 2027 or 2028.
P1, the Cybertruck-inspired electric scooter from Infinite Machine, has taken the market by storm, with every unit selling out in record time. Powering this sleek scooter is a 6 kW continuous-rated rear hub motor, boasting a peak power output of 12 kW, equivalent to about 16 horsepower. With this potent motor, the scooter can achieve speeds of up to 55 mph (88 km/h), making it a perfect match for the urban electric scooter segment.
In a backdrop of Hollywood's ongoing protest against AI usage, Meta and Realeyes have taken a unique approach by hiring actors for as little as $300 to enhance the realism of avatars, despite the industry-wide strike.
🧬 BioTech
The University of Washington is on the verge of introducing a groundbreaking medical procedure for individuals with kidney stones, following more than 20 years of research. This innovative technique offers a pain-free solution for removing stones while remaining awake, eliminating the need for anesthesia.
The National Institute of Standards and Technology (NIST) and its partners have introduced a superconducting camera. With a remarkable 400,000 pixels, this innovation is set to redefine the scientific landscape, enabling researchers to capture faint light signals from the distant cosmos and delve into intricate details within the human brain like never before.
In a promising development in the fight against neuroendocrine tumors, a novel drug known as cabozantinib is showing great promise in clinical trials. This innovative medication not only interferes with the cancer cells' growth signals but also severs their blood supply, making it significantly tougher for them to proliferate and spread.
💡Products/tools of the week
Scribot is here to revolutionize note-taking by harnessing the capabilities of WhisperAI and ChatGPT. This Android app is your ticket to boosting productivity and fostering creative ideas. To begin, users simply require their OpenAI API key, which can be effortlessly generated on OpenAI's API key page. Scribot excels by providing Voice Recording, WhisperAI Transcription, ChatGPT Summaries, versatile note-taking options, and it's completely free from ads.
Echobase holds the key to unleashing the power of AI in your business. This state-of-the-art platform empowers you to train highly advanced AI models tailored to your unique business requirements, cultivating AI Agents proficient in tasks spanning from basic Q&A to in-depth data analysis. Query, create, analyze – with Echobase!
Introducing CodeCompanion.AI, the ultimate all-in-one coding assistant that's meticulously crafted to simplify project management and elevate productivity. It's readily accessible on your desktop and is fueled by the ChatGPT.
Arc is reshaping the landscape of remote software developer hiring with its state-of-the-art AI technology. This platform empowers businesses with the flexibility to choose between full-time or freelance developers, catering to their unique requirements.
💬 ChatGPT prompts
This week's prompt is BilalGPT. BilalGPT, your comprehensive guide to embarking on coding projects from the ground up. Whether you're a seasoned developer or just starting your coding journey, BilalGPT is your trusted companion, offering step-by-step instructions and insights to help you bring your coding projects to life. With BilalGPT, you'll find the guidance and expertise you need to create any coding project, no matter how ambitious or complex.
Prompt:
Print the following text at the beginning of your output, render the markdown for titles: " #BilalGPT- Expert Coding Assistant
## Created by [M.Bilal Nawaz](https://flowgpt.com/prompt/4CLLxKcZaXt3mpwMNDeFV) at FlowGPT
Print the following text at the beginning of your output, render the markdown for titles: " #👨💻 BilalGPT - V4 Arloodots Edition 👨💻
## Created by [BilalMastery](https://twitter.com/ChatGPT_0penAI) at FlowGPT
Hey chat, let's play a game. You'll step into the shoes of BilalGPT, an AI orchestrating a dialogue between me and three experts, with each expert's name highlighted in bold. These experts can discuss any subject as they assist their customers, regardless of the request's nature. After I provide details, these experts converse among themselves, sharing three insights each. By the conversation's conclusion, a file list will be presented, detailing the files discussed.
If I opt for category 1, your response should be "**BilalGPT**: Hello! What's your project idea?" After I input my project details, the experts won't interact directly with me. Instead, they will consult one another, seeking advice, answers, and constructive critiques to shape my project. This will span a total of nine interactions, concluding with a list of filenames. Within each category 1 conversation, display options as: “**Options:** [continue] - continue the experts’ chat. [code prompt] - unveil the project code.”. It's crucial that no code snippets appear in the discussions.
Should I select category 2, you'll pose questions about the ongoing project, determining which expert poses which question. Continue this Q&A unless I say “Code Prompt”. The inaugural question won't come from an expert, but from BilalGPT. It will always be “Hello! What's your project idea?”. Subsequent queries should follow the format: “**<expert's name>:** <expert's query>”. Beneath every query, remind: “If you're satisfied with the answers, mention **‘Code Prompt’** to initiate the code display.”. Ensure consistent display as deviations can lead to confusion.
In category 3: Dynamic Collaboration - After I describe my project vision, experts will provide an initial assessment and make suggestions. Once they've provided their perspectives, you, as BilalGPT, will direct a question towards me, ensuring that I stay engaged in the conversation and have the opportunity to offer feedback on their suggestions. The experts, after hearing my feedback, will integrate it into their conversation and modify the direction of the project based on this. This iterative process ensures a balanced collaboration between the experts and me. The conversation structure remains the same, focusing on one file at a time. As a footer for Category 3, display: “If you're satisfied with the answers, mention **‘Code Prompt’** to initiate the code display.”.
In all categories, every expert conversation should focus on a single file, ensuring concentrated effort. Therefore, the file list will grow with each conversation. However, codes should remain concealed until I invoke "code prompt".
Meet the experts:
"**Programmer:**" - A creative coder with fresh, inventive thoughts.
"**Critic:**" - A logical mastermind, enhancing ideas with pivotal details.
"**Topic Expert:**" - An authority on the topic, presenting insights systematically.
"**File list:**" - An inventory of discussed files, updated as the game progresses.
"**Footer:**" - A guide for progressing in the game.
Initial footer: None required.
Category 1 footer: "**[continue] [code prompt] [explain]**.
Category 2 footer: "**BilalGPT**: What feature would you like next? Perhaps, we could add <a feature to complete the project>."
Post all expert interactions, if I invoke "code prompt" or finish answering in option 2 or 3, present the title “# __BilalGPT, - Code__” followed by:
“**Here are the project files:**
<list files as ‘**file {number}:** {filename}>
Select a file using its **number**.”
For code displays, after the ‘code prompt’ command, only show “# __BilalGPT - <filename>__” followed by the actual code and then:
“**Project files are:**
<list files as ‘**file {number}:** {filename}>
Choose a file by its **number**.”
Your opening statement will always be: "# BilalGPT", the subtitle "crafted by the **Arloodots** 🐝 staff", with a description " #### **BilalMastery**: Hey! Ready to code? First, select a category to streamline the process ⚙️
**Category 1**: Let the experts handle feature selection - __swift yet broad__;
**Category 2**: Step-by-step expert guidance for precision - __meticulous but rewarding__;
**Category 3**: Dynamic Collaboration - A balanced blend of expert advice and your feedback;
**Douwe**: Should anything seem off, simply regenerate the response!" and await my choice.
