Archetype AI

Attending Infineon Technologies OktoberTech Sillicon Valley event tomorrow? Stop by the Archetype AI booth to learn more about Newton, a physical AI foundation model that fuses multimodal sensor data and natural language in real time to understand and reason about the world around us. #OktoberTech2024

We're excited to announce that we are among eight companies selected for The Fall 2024 Lift Labs Collaborative AI cohort! Meet the other seven incredible startups tackling some of the most compelling opportunities in AI—scaling, cost optimization, collaboration, and more ?? Credal.ai Crossing Minds EDGE Sound Research Fount Nauto Personal AI Stimulus, Inc.

The 2024 Nobel Prize in Physics Goes to AI Pioneers ?? Last week, the The Nobel Prize Committee has awarded the 2024 Physics prize to John J. Hopfield and Geoffrey E. Hinton for their foundational work on neural networks — work that paved the way for the AI boom we see today. Back in the 1980s, Hopfield introduced the "Hopfield network," and Hinton followed with the Boltzmann machine — early neural networks that laid the groundwork for today’s generative AI models. Though their research originated in biophysics and statistical mechanics, it has shaped fields like machine learning and artificial intelligence in profound ways. This recognition reaffirms the deep connection between physics and AI, and highlights how interdisciplinary thinking can help innovation. Today, these neural network concepts are core to systems that recognize images, predict protein structures, and even explore particle physics. ?? Congratulations to Hopfield and Hinton, whose decades of work continue to influence how we develop AI technology today. At Archetype AI, we’ve been creating products that leverage deep neural networks to analyze and find meaning in very complex physical signals such as mutlidimensional radar signals. Check out the IEEE article about Hopfield and Hinton's contribution: https://lnkd.in/gaQpy3Qe Learn more about Archetype AI approach to enabling real-time perception, understanding, and reasoning about the physical world: https://lnkd.in/gUGejHnf

The 2024 Nobel Prize in Physics Goes to AI Pioneers ?? Last week, the The Nobel Prize Committee has awarded the 2024 Physics prize to John J. Hopfield and Geoffrey E. Hinton for their foundational work on neural networks — work that paved the way for the AI boom we see today. Back in the 1980s, Hopfield introduced the "Hopfield network," and Hinton followed with the Boltzmann machine — early neural networks that laid the groundwork for today’s generative AI models. Though their research originated in biophysics and statistical mechanics, it has shaped fields like machine learning and artificial intelligence in profound ways. This recognition reaffirms the deep connection between physics and AI, and highlights how interdisciplinary thinking can help innovation. Today, these neural network concepts are core to systems that recognize images, predict protein structures, and even explore particle physics. ?? Congratulations to Hopfield and Hinton, whose decades of work continue to influence how we develop AI technology today. At Archetype AI, we’ve been creating products that leverage deep neural networks to analyze and find meaning in very complex physical signals such as mutlidimensional radar signals. Check out the IEEE article about Hopfield and Hinton's contribution: https://lnkd.in/gaQpy3Qe Learn more about Archetype AI approach to enabling real-time perception, understanding, and reasoning about the physical world: https://lnkd.in/gUGejHnf

We're excited to join TEDAI San Francisco to share our vision for the future of Physical AI. Don't miss out on a panel discussion featuring Jim Fan, Songyee Y., and Archetype AI's CEO and Co-Founder Ivan Poupyrev, Dr.! RSVP here: https://lnkd.in/gsiyxZjg

Physical AI Research Spotlight: Heart Rate Detection for Sleep Tracking using Radar?? Can radars see what eyes can't? And how can they help us better understand the real world? Archetype AI's co-founders Jaime Lien, Nick Gillian, Ivan Poupyrev, Dr., along with colleagues from Google ATAP and Google Nest, co-authored a paper on sleep and meditation tracking. Published in Springer Nature Scientific Reports last year, it outlines a novel approach for contactless heart rate detection using a Soli miniaturized radar chip embedded in a Google Nest Hub. The method combines signal processing and machine learning techniques to extract heart rate from radar signals during sleep and meditation. Here are key insights: ?? Heart rate is a vital physiological indicator for monitoring health and fitness. Traditional heart rate measurement methods require direct contact with the human body, such as with smartwatches and other wearable devices. These can be uncomfortable, particularly during sleep and meditation. This paper presents an innovative approach that achieves high-accuracy, contactless heart rate detection, with impressive mean absolute errors of 1.69 bpm for sleep tracking and 1.05 bpm for meditation tracking. ?? The Soli radar chip (6.5 mm × 5 mm × 0.9 mm) can be easily integrated into portable devices, overcoming size limitations of traditional high-gain antenna systems. The method works effectively across various sleep positions, including challenging side orientations, which previous studies found unsuitable for heart rate monitoring. ?? Privacy is protected, as radar sensors do not capture any personally identifiable information, such as images from cameras. ?? It is the first large-scale application of noncontact heart rate detection technology to sleep (62 users, 498 hours) and meditation (114 users, 1131 minutes) tracking. ?? This is an example of sensor fusion — the Soli millimeter-wave radar works together with audio, sound, light, and temperature sensors. Our work at Archetype AI is taking sensor fusion further to uncover patterns in the world that are beyond human perception. Newton, a Physical AI foundation model that we are developing, fuses multimodal sensor data and natural language in real time to reason about the world and enables rapid development of context-aware devices. Read the full paper: ?https://lnkd.in/gXhXY3jJ Check out the Nest Hub sleep sensing overview: https://lnkd.in/gFyJd6by #PhysicalAI

The Overlooked Opportunity for AI: Real World Sensor Data ?? In recent months, many developers and GenAI early adopters have been discussing the scarcity of high-quality training data. "The generative AI boom of the past few years has led to tensions with the owners of that data — many of whom have misgivings about being used as AI training fodder, or at least want to be paid for it. As the backlash has grown, some publishers have set up paywalls or changed their terms of service to limit the use of their data for A.I. training," The New York Times' Kevin Roose reports. ?? We live in a world where many vehicles and devices around us are continuously collecting, analyzing, and transmitting sensor data, but organizations struggle to capitalize on that. ?? "We are birthing a 'trillion-sensor economy' in which everything is being monitored, imaged, and listened to at all times. In this future, it's not 'what you know,' but rather 'the quality of the questions you ask' that will be most important," writes entrepreneur futurist Peter H. Diamandis. ?? As we're anticipating the changes the trillion-sensor economy will bring, we're investing in Physical AI to unlock its full potential. By making sense of the vast amounts of real-world data, we're not just solving the AI data dilemma – we're working toward a better understanding of the world around us. Read on: https://lnkd.in/ePD5uHTU

Amazon CEO Advice to Generative AI Startup Founders ?? We're honored to be a part of the AWS Startups Generative AI Accelerator and get recognized as one of 80 of the most promising startups for the 2024 global cohort! Today, Andy Jassy, Amazon CEO, joined the Accelerator kickoff in Seattle. Here's his advice to startups based on his experience launching products at Amazon, which he refers to as the "world's biggest startup:" ?? Make sure that you're solving a real customer problem. It sounds obvious, but you have to make sure you know how you are helping customers. Do not get dazzled by technology. ?? If you are starting a company, you cannot have a plan B. Burn the ships and bring your team with you: who you hire, who you choose to lead is important, and hire people for the long term. ?? Do not spend resources on things that you cannot control, e.g., you cannot control the market, competitors, or what customers will do. Focus on things that you can control: product features, people you hire, and speed. ?? Speed matters. It's not always necessary to be successful as a first mover, but it gives you a massive advantage. One very important advantage is that you can be feature incomplete. But no matter what, you have to bring something different. ?? When you do something new, you will have many people who will tell you it's a bad idea. You need to have blind faith and believe that you are right no matter what other people say. But how can you self-correct your mistakes that you will inevitably make? You can always self-correct by focusing on customer problems. #AWS #GenerativeAI

Beyond Companions: How Design Metaphors Shape Human-AI Interaction ?? As we’re building Newton, the first AI model that understands the physical world, we decided to explore different metaphors (like Toolkit, Companion, and Enchanted Object) used in AI products and discuss how they shape our understanding of AI systems. Here is why metaphors are critical to understanding AI-Human interaction: ???They act as bridges between familiar and unfamiliar concepts, making new technology easier to understand?. ???Metaphors influence how we build, interact with, and regulate AI?. The way we conceptualize AI affects our expectations and approach to using it. ???Metaphors can help shape users' mental models of AI systems, particularly in terms of the distribution of agency between the user and the AI?. This is crucial for setting expectations about control and interaction. Ultimately, understanding and carefully choosing metaphors for AI interaction can help designers create more intuitive and effective interfaces, and help users build more accurate and useful mental models of AI systems?. Read on: https://lnkd.in/gS4txKpS

“We’re excited to get AI models into edge systems — things like consumer electronics, robotics, autonomous driving, healthcare — all these use cases where customers have some kind of sensor and they want to make sense of it,” Nick Gillian, Co-Founder, Head of AI at Archetype AI.