Inteligencia Artificial (IA)
Innovations in AI: ChatGPT in Health, Advanced Robots, and Autonomous Vehicles
Paloma Firgaira
2026-01-11
5 min read
The week has been dominated by CES in Las Vegas, the most influential technology event globally. This edition has seen Nvidia and robotics take center stage, along with innovations in smart glasses and other devices. Humanoid robots are emerging as the stars of the year: their hardware and skills have advanced significantly, and while they still do not match human speed, they are close to achieving it. The sector anticipates a "ChatGPT moment" for robotics, meaning the arrival of AI capable of giving robots true autonomy in real-world environments.
Will 2026 be the key year? Many experts believe so. In the meantime, robots are already being integrated into factories, hospitals, hotels, and logistics. One of the most anticipated innovations has been ChatGPT Health, the healthcare version of ChatGPT, which, although not yet available in Europe, reinforces OpenAI's commitment to making its AI a comprehensive personal assistant.
Nvidia has introduced its new Rubin processor, four times more powerful than its predecessor, with lower energy consumption and a more efficient design. This advancement marks a paradigm shift: inference, or the execution of AI models, is gaining prominence over training. In this context, the acquisition of Groq, which specializes in inference chips, strengthens Nvidia's strategy.
Nvidia's announcements have focused on two areas: autonomous vehicles and robotics. In the automotive sector, they presented Alpamayo, a reasoning model for autonomous cars, robotaxis, and trucks, which interprets the environment rather than just reacting. While it does not reach the level of Tesla or Waymo, it offers other manufacturers a competitive foundation. Brands like Mercedes, Stellantis, and BYD are already testing it.
For Nvidia, this represents a new business avenue, both in hardware and software. Elon Musk has noted that Alpamayo follows Tesla's approach, solving most situations, although exceptional cases remain a challenge.
In robotics, Nvidia has launched the Cosmos and Isaac models, which allow training robots in simulations before deploying them in the real world. Boston Dynamics, now under Hyundai, has introduced the new Atlas, a robot with superior mobility to humans, 360° vision, and integration with Google DeepMind, which will initially be implemented in Kia factories.
Samsung showcased a 3D television and a foldable tablet; Asus presented glasses similar to Meta's; Razer introduced a personal assistant; and Lucid unveiled an adaptable robotaxi for any car, which Uber will incorporate into its fleet. Additionally, there is an abundance of multifunctional Chinese robots.
ChatGPT aims to become a comprehensive assistant, including health. ChatGPT Health will feature a secure area for medical data, integrating analytics and wearable data. With over 230 million weekly users on health topics, this function, developed alongside medical professionals, could transform primary care and patient monitoring, with plans for integration into public and private healthcare systems.
In the energy sector, Finnish startup Donut Lab has presented a solid-state battery, considered the "holy grail" of the industry. It promises 400 Wh/kg density, charges in 5 minutes, and has a lifespan of one million cycles, functioning between -30 °C and 100 °C, and made with affordable materials. Ultra Bikes is already using these batteries, achieving ranges of up to 600 km. However, skepticism remains about the real viability of these claims.
In AI research, DeepSeek's work on Manifold-Constrained Hyper-Connections (mHC) stands out, a technique that improves the training of large language models by preserving original information and increasing stability and scalability.
From Stanford, the SleepFM model, trained with over 585,000 hours of sleep records, can predict the risk of 130 diseases from just one night of data, surpassing traditional models in accuracy and applicability.
Meta has proposed training AI "co-scientists" using rubrics extracted from scientific articles, allowing AI to generate and evaluate more rigorous research plans through reinforcement learning.
In the business landscape, Google has integrated Gemini into Gmail, enhancing text correction and assistant functions, raising concerns for Grammarly. Anthropic has raised $10 billion and reached a valuation of $350 billion, expecting to triple its revenue this year. Google has surpassed Apple as the second most valuable company in the world. Nvidia requires upfront payment from Chinese companies for its processors due to regulatory uncertainty. OpenAI reserves $50 billion for stock options, 10% of its value. xAI reaches a valuation of $230 billion, driven by Nvidia. LMArena closes an investment round, reaching $1.7 billion. In Utah, an AI system that renews medical prescriptions without human intervention has already been approved.