sonyai.bsky.social @sonyai.bsky.social · 01/07/2026Five years ago we asked: can an AI agent outrace the world's best #GranTurismo drivers? The answer became a @nature.com cover, a game feature and a research frontier that's still open. #GTSophy, five years on ↓ bit.ly/4vHyxpF 011
sonyai.bsky.social @sonyai.bsky.social · 01/07/2026June at Sony AI: our robot #Ace beat seven ranked pro table tennis players under official rules, and the #robotics press is taking note. 🏓 We brought six papers to #CVPR 2026 in Denver. Alice Xiang made the case for data rights in @time.com. Plus our work at #FAccT 2026. Catch up: bit.ly/4wh2aOd 000
sonyai.bsky.social @sonyai.bsky.social · 10/06/2026Since our @nature.com paper, #Ace won against seven pro table tennis players, including two-time Olympic silver medalist Miu Hirano, and world No. 26 Miyuu Kihara. Learn how improvements across #RL, hardware, #perception, and more helped close the gap: bit.ly/440oyPM 000
sonyai.bsky.social @sonyai.bsky.social · 01/06/2026Sony AI is presenting 10 papers at #CVPR2026 in Denver, covering generative modeling, face aging, video-to-audio synthesis, 3D scene understanding, and evolving-world perception. 👉 bit.ly/4obvwuD 020
sonyai.bsky.social @sonyai.bsky.social · 13/05/2026#SPARC: a single racing policy that holds up across 100+ unseen vehicles — no vehicle-specific tuning needed. 👉 Learn more: arxiv.org/pdf/2511.09737 011
sonyai.bsky.social @sonyai.bsky.social · 06/05/2026 👉 bit.ly/4wjSmE8 #SonyAI #AIGovernance #AIEthics 020
sonyai.bsky.social @sonyai.bsky.social · 06/05/2026@alicex84.bsky.social, Global Head of AI Governance at Sony and Lead Research Scientist @sonyai.bsky.social, writes in @time.com: data nihilism is driving one of the greatest wealth transfers in history. The solution is ethical innovation built on consent and shared value.bit.lyAre We Entering the Age of Data Nihilism?People feel powerless to control their data, which could widen inequality in the AI era warns Alice Xiang. 111
sonyai.bsky.social @sonyai.bsky.social · 06/05/2026#ICASSP2026 Sony AI is presenting 11 papers in Barcelona this May, covering #music understanding, #generativeaudio, audio-visual alignment, and #dataquality. Each addresses a different gap between what #audio AI systems produce and what they actually understand. 👉 Learn more: bit.ly/4dsZPsX 050
sonyai.bsky.social @sonyai.bsky.social · 05/05/2026Lorenzo Servadei, Head of AI for Chip Design at Sony AI, speaks with Silicon Semiconductor on AI-powered EDA, #GENIE-ASI for analog subcircuit identification, and #Schemato for human-readable schematics. Read the interview: siliconsemiconductor.net/magazine/686 #SonyAI #ChipDesign 020
sonyai.bsky.social @sonyai.bsky.social · 30/04/2026A defining month for Sony AI: Project #Ace reached the cover of @nature.com, a new framework on #AI companionship, a full slate at #ICLR 2026, and a feature on the future of #chip design. 🔗 Read the April Month in Review: bit.ly/3OEuutX 010
sonyai.bsky.social @sonyai.bsky.social · 27/04/2026Watch the full Project Ace short film, dive into research and team interviews at: ace.ai.sony #SonyAI #PhysicalAI #Robotics #TableTennis #AIResearch 000
sonyai.bsky.social @sonyai.bsky.social · 27/04/2026Second: safety. From day one, the team worked with safety professionals through a formal certification process. Researchers started in helmets, pads, and glasses. The system is now operating at competition speed in direct physical interaction with humans—a first. 100
sonyai.bsky.social @sonyai.bsky.social · 27/04/2026First: perception. Ace is the only system in the world capable of measuring spin on an unaltered table tennis ball at competition speed. That capability underpins everything else the system does. 100
sonyai.bsky.social @sonyai.bsky.social · 27/04/2026AI has largely existed in virtual spaces. Ace was built to change that—and today, that research was published in @nature.com. Michael Spranger, President of Sony AI, explains what it actually required. 100
sonyai.bsky.social @sonyai.bsky.social · 27/04/2026👉 Dive into the research and more at ace.ai.sony #SonyAI #Robotics #TableTennis #AIResearch #MachineLearning 000
sonyai.bsky.social @sonyai.bsky.social · 27/04/2026That's why Ace—Sony AI's research, published in @nature.com—is built on reinforcement learning. A system that develops its own responses to situations it has never encountered before. 100
sonyai.bsky.social @sonyai.bsky.social · 27/04/2026Rule-based systems don't scale. Specifying racket angles for every spin type, anticipating every shot a human might play it's an impossible list. And human opponents don't stay predictable: they adapt mid-match, actively exploiting any weakness they find. 100
sonyai.bsky.social @sonyai.bsky.social · 27/04/2026Why can't you just program a robot to play table tennis? Peter Dürr and Peter Stone explain, and the answer gets to the heart of what makes physical AI hard. 100
sonyai.bsky.social @sonyai.bsky.social · 24/04/2026Over several years, we’ve contributed #research to @iclr-conf.bsky.social exploring how #machinelearning models are trained, interpreted, and applied. This year’s papers span #multimodallearning, #diffusion, interpretability, and theory, with open #code and demos. 🔗 bit.ly/4eFumVs 162
sonyai.bsky.social @sonyai.bsky.social · 24/04/2026Watch the video to see how Ace operates under uncertainty, and visit ace.ai.sony to explore the research. #SonyAI #AIResearch #Robotics #TableTennis #MachineLearning #Ace #ReinforcementLearning 000
sonyai.bsky.social @sonyai.bsky.social · 24/04/2026In experiments, Ace reacted within ~49 milliseconds after net contact, adapting its motion and successfully returning the ball even in these unexpected situations. 100
sonyai.bsky.social @sonyai.bsky.social · 24/04/2026Ace responds by simulating multiple possible ball trajectories in real time. The green overlays in the video show alternative paths the ball could take, allowing the system to adjust its motion quickly after the bounce. 100
sonyai.bsky.social @sonyai.bsky.social · 24/04/2026What happens when a ball hits the net? In table tennis, net contact creates unpredictable trajectories. For Ace, Sony AI's physical AI research system, these rare events were one of the hardest real-world conditions to address. 162
sonyai.bsky.social @sonyai.bsky.social · 24/04/2026Watch the video to see how Ace plans and commits to serves in real time. Visit ace.ai.sony to learn more about the research. #SonyAI #AIResearch #Robotics #TableTennis #MachineLearning #Ace 000
sonyai.bsky.social @sonyai.bsky.social · 24/04/2026Across matches, Ace executed multiple serve types designed to land in different areas and produce different spins, making it harder for opponents to anticipate the next shot. 200
sonyai.bsky.social @sonyai.bsky.social · 24/04/2026Ace draws from a library of serve motions. These serves vary in spin, speed, and landing position, allowing the system to select different strategies and introduce variation during matches. 100
sonyai.bsky.social @sonyai.bsky.social · 24/04/2026How does Ace—Sony AI's robotic AI research system—plan a serve? Published in @nature.com, Ace is the first robot to beat a professional athlete in a physical sport. Before a serve, Ace predicts how spin will affect the ball's flight and bounce to guide real-time execution. 110
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026Watch the video to see how the Gaze Control System works and visit bit.ly/4mNu1Cf to explore behind-the-scenes. #SonyAI #AIResearch #Robotics #TableTennis #ComputerVision #Ace #ReinforcementLearning #ArtificalIntelligence 010
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026Understanding spin is essential to returning high-level shots. Ace integrates this perception system with its control framework to respond during live rallies. 120
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026To do this, Ace combines event-based vision sensors, mirror-based tracking, and a telephoto lens to keep the ball in focus while estimating its angular velocity. These measurements allow the system to infer spin and speed hundreds of times per second during play. 120
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026How does Ace read spin in real time? Published in @nature.com, Ace is the first robot to beat a professional athlete in a physical sport. Its Gaze Control System tracks spin during fast rallies, helping it interpret the game as it unfolds. 1215
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026The film goes behind how it was built: the hardware, the breakthroughs, and the moments the team wasn't sure it was going to work. Watch it now: bit.ly/48lxH8f For the research + interviews: bit.ly/4mNu1Cf #SonyAI #PhysicalAI #Robotics #TableTennis #AIResearch 000
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026Ace is Sony AI's robot athlete, and the subject of the research project published in @nature.com. It tracks a spinning ball at up to 700 Hz, responds with an end-to-end latency of 20.2 milliseconds, and competes against elite and professional players under official rules. 100
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026The first robot ping-pong prototype appeared in 1983. For more than forty years, no machine could see, decide, and act within the window required to compete with elite players. That changes today. 110
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026This video breaks down how those three components come together and why getting the integration right was the central engineering challenge of the project. For the full research and team interviews: bit.ly/4mNu1Cf #SonyAI #PhysicalAI #Robotics #TableTennis #AIResearch 000
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026For Ace, Sony AI's table tennis robot, that integration happens in 20.2 milliseconds end-to-end. 9 cameras, three event-based vision sensors, a custom-built robotic arm, and a reinforcement learning policy that was trained entirely in simulation before ever touching a real table. 100
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026Every autonomous robot depends on three functions: sensing, deciding, and acting. The challenge isn't building each one. It's integrating all three fast enough to be useful. 100
sonyai.bsky.social @sonyai.bsky.social · 23/04/2026Sense. Decide. Act. In milliseconds. This is the story of a robot that can beat professional table tennis players. Watch the short film: ace.ai.sony #SonyAI #ACE #Robotics 010
sonyai.bsky.social @sonyai.bsky.social · 22/04/2026Read the full story and explore the research at ace.ai.sony. Press Release: bit.ly/4csxWB2 Feature Blog: bit.ly/4mNu1Cf #SonyAI #AIResearch #PhysicalAI #Robotics #MachineLearning #TableTennis 000
sonyai.bsky.social @sonyai.bsky.social · 22/04/2026This is a research milestone with implications for physical AI, assistive robotics, and how machines interact with humans in high-speed environments. 100
sonyai.bsky.social @sonyai.bsky.social · 22/04/2026As Sony AI Chief Scientist, Peter Stone, explains: "This is bigger than table tennis. It's the very first time there's been a human expert-level demonstration of competitive play in the real world across any sport." 100
sonyai.bsky.social @sonyai.bsky.social · 22/04/2026It combines event-based vision (tracking the ball at up to 700Hz), a custom 8-DOF robotic arm, and deep #reinforcementlearning transferred from simulation to the real world, with end-to-end latency of 20.2ms. 100
sonyai.bsky.social @sonyai.bsky.social · 22/04/2026Ace is the first AI system to compete with and beat elite and professional players under official rules. 100
sonyai.bsky.social @sonyai.bsky.social · 22/04/2026For 40+ years, building a robot that could rally with an elite human table tennis player at full speed was an unsolved problem. Sony AI's Ace research project set out to change that—and the results are now accepted for publication in @nature.com and featured on the cover. 141
sonyai.bsky.social @sonyai.bsky.social · 02/04/2026March at Sony AI: Jesse Lai on writing The Principles of Diffusion Models, 11 papers accepted to ICASSP 2026, and Alice Xiang on the Me, Myself and AI podcast discussing FHIBE. Full recap: bit.ly/4c4oBxO #SonyAI #AIResearch #MachineLearning #ICASSP2026 000
sonyai.bsky.social @sonyai.bsky.social · 31/03/2026SHIELD benchmarks how #vision-language models detect #AI-edited images without task-specific training. 👉 Read more: bit.ly/4dRuHE3 000
sonyai.bsky.social @sonyai.bsky.social · 24/03/2026EnTruth offers a new way to trace unauthorized dataset use in #diffusionmodels with minimal image alteration. 👉 Full paper: bit.ly/4sXLlpO 000
sonyai.bsky.social @sonyai.bsky.social · 11/03/2026StelLA shows how structured low-rank adaptation can outperform standard #LoRA without added compute. 👉 Read the paper: bit.ly/46PkK62 000
sonyai.bsky.social @sonyai.bsky.social · 03/03/2026Music Arena brings live, human preference evaluation to text-to-music #research. 👉 Read more: bit.ly/4aMmeQH 000
sonyai.bsky.social @sonyai.bsky.social · 26/02/2026FlashFoley enables real-time, controllable sketch-to-audio generation. 👉 Read the paper: bit.ly/46qQQVn 000
