šļø Computer Vision is the Main Topic at NeurIPS 2025
Plus: The Real AI Impact on jobs is way Bigger Than We Think
Hope everyone is well. Iām reporting from NeurIPS Mexico City, where hundreds of researchers and professors are gathering to share their research and latest work on AI. Computer Vision has been the main topic of the conference, highlighting the importance that this tech has on AV, robotics, and medicine. I will share more. Also, we share how AI is impacting Jobs worldwide, which is ~5x higher than we think. Letās dive in and stay curious.
Computer Vision is the Main Topic at NeurIPS 2025
š§° AI Tools - Event Planning
The Real AI Impact on jobs is 5Ć Bigger Than We Think
š ļø AI Jobs Corner
š AI Learning Resource - AI Conferences
Subscribe today and get 60% off for a year, free access to our 1,500+ AI tools database, and a complimentary 30-minute personalized consulting session to help you supercharge your AI strategy. Act now as it expires in 3 daysā¦
š° AI News and Trends
Can US AI models compete with Chinese Models? DeepSeekās V3.2 matches GPT-5, and its higher-compute V3.2-Speciale variant rivals Gemini-3.0-Pro and earned gold medals at IMO, IOI, and ICPC 2025. This is why OpenAI has issued a ācode redā warning to its employees. āImprove ChatGPT or you will be firedā
OpenAI Takes Stake in Thrive Holdings, a Buyer of Services Firms
ByteDanceās Doubao app is Chinaās most popular mobile AI platform. With ~11.4 million downloads in October, Doubaoās focus is on frictionless AI-powered voice, image, and video.
A former Microsoft AI VP is replacing John Giannandrea as Appleās head of AI.
How a professor is benefiting from AI use in the classroom
Other Tech News
India demands Apple preinstall government security app onto iPhones
Neuralinkās first patient could receive an upgrade allowing the patient to do more than play video games, attend neuroscience classes, and work, which are the current capabilities.
Samsung has revealed its first tri-folding phone ahead of Apple
Shopify Breaks Down on Busy Cyber Monday
Computer Vision is the Main Topic at NeurIPS 2025
This past Sunday, November 30th, I had the pleasure of facilitating my second LXAI workshop at NeurIPS 2025 in CDMX. NeurIPS, Neural Information Processing Systems (formerly NIPS), is a machine learning and computational neuroscience conference held every December. Along with ICLR and ICML, it is one of the primary conferences of high impact in machine learning and artificial intelligence research. The conference usually takes place in one centralized location, but this year, due to the US immigration situation, it was split between San Diego and Mexico City. The highlight of this year's conference, I feel it was Computer Vision.
Most of the talks focused on newer technologies to train models utilizing images, videos, and 3D environments to improve output and to also find ways to save on energy and computer power usage, and streamline these training processes. As every December, thousands of researchers, professors, and students, as well as entrepreneurs from all corners of the world, meet at NeurIPS to share their work, network, and start new ventures and research.
Nvidia was not left behind, obviously, unveiling 70+ papers and a powerful set of open-source tools across digital and physical AI. Highlights include
Alpamayo-R1, the first open reasoning VLA model for autonomous driving.
New Cosmos tools like LidarGen and ProtoMotions3 for robotics and simulation.
And expanded Nemotron models for speech, safety, and synthetic data already adopted by partners like CrowdStrike, Palantir, and ServiceNow.
NVIDIA also earned top marks for openness from the new Artificial Analysis Openness Index. NeurIPS matters because itās where breakthroughs launch, where top researchers stress-test ideas, and where the direction of AI is set for the year. On the ground this year, one trend dominates: computer vision everywhere, powering everything from AVs to robotics to multimodal reasoning.
Both NeurIPS conference locations are strong right now until the end of the week. Being here allows my team and me to dive into the future of AI, talk to researchers who are driving this technology, and network to remain on top of it. I will continue to share more with you.
š ļø AI Jobs Corner
Apply Today - Open Positions.
First-Line Supervisors of Productions and Operating Workers
Computer and Information Systems Managers
The Real AI Impact on jobs is 5Ć Bigger Than We Think
Project Iceberg, a recent study by MIT, introduces the Iceberg Index, a new KPI that measures how much of each jobās wage value comes from skills AI can already perform, instead of waiting for layoffs or unemployment data.
Using 151M workers, 923 occupations, 3,000 counties, 32k skills and 13k+ AI tools, the team simulates the U.S. labor market on the Frontier supercomputer and finds that the visible AI disruption in tech is just the tip of a much larger iceberg: current AI adoption in computing and tech jobs accounts for only 2.2% of total wage value (~$211B), while technical capability across admin, finance and professional services reaches 11.7% (~$1.2T), about 5Ć larger and spread across every state, not just coastal hubs. Traditional metrics like GDP, income, and unemployment explain under 5% of this variation, meaning policymakers and businesses relying only on standard stats are missing where AI exposure really is. The Index doesnāt predict job losses; it maps where human and AI skills overlap, so leaders can simulate scenarios, spot āautomation surpriseā regions (e.g., Rust Belt states with big hidden white-collar exposure), and target training, infrastructure, and reskilling before disruption shows up in the data.
Stats to keep in mind:
Scale: AI is already technically capable of ~16% of classified labor tasks.
Exposure gap: Tech-sector disruption = 2.2% wage value vs. hidden cognitive/admin exposure = 11.7% (~$1.2T).
Coverage: Model includes 151M workers, 32k skills, 13k+ AI tools across 3,000 counties.
Validation: Skill embeddings predict 85% of real career transitions; state exposure aligns 69% with Anthropicās AI usage index.
Policy use: Iceberg Index acts like a risk map for AI exposure, letting governments and firms test workforce strategies in simulation instead of reacting after the fact.
AI Learning Corner
Largest AI Conferences
AI & Big Data Expo: A series of events globally, focusing on AI, machine learning, and big data in a business context.
AAAI Conference on Artificial Intelligence: A primary academic and research conference for artificial intelligence.
World Summit AI: A large event focused on the business applications of AI and strategy.
Ai4: A large conference in North America focused on the practical application of AI across various industries.
AWS re:Invent: Amazon Web Servicesā large annual event with a significant focus on AI and cloud computing.
NVIDIA GTC: The GPU Technology Conference, a key event for those interested in GPU-accelerated computing and AI.
International Conference on Machine Learning (ICML): A premier academic conference for machine learning research.
ICLR: The premier gathering of professionals dedicated to the advancement of representation learning aka deep learning.
š§° AI Tools of The Day
Event Planning
Bizzabo ā AI-powered event intelligence platform that helps manage large and corporate-scale events. It tracks attendee engagement in real time, generates analytics, and helps prove event ROI.
Grip ā Designed for conferences, trade shows, and hybrid events. It offers AI-driven matchmaking & networking, personalized schedules, exhibitor directories, badge scanning, and more ā helping manage exhibitor/attendee interactions and logistics at scale.
Zkipster ā Good for guest-list management, check-ins, seating, and RSVP tracking. For large events, Zkipsterās automation & guest-flow tools make check-in smoother and seating/guest-management much more efficient.
Outstanding coverage from NeurIPS! Your observation about computer vision being the dominant theme really captures the shift happening in AI research right now.
What strikes me most is how the field is moving beyond just accuracy gains toward energy-efficient training methodologies. The emphasis on streamlining compute usage while handling images, video, and 3D environments suggests we're entering a maturation phase where practical deployment constraints are shaping research priorities as much as benchmark performance. NVIDIA's Alpamayo-R1 for autonomous driving is a perfect exaple of this trend, combining reasoning capabilities with vision in a way that's actually deployable.
This practicality focus might be what ultimately bridges the gap between conference research and real-world robotics and medical applications you mentioned.