As a child, Luis Eduardo Garza Elizondo couldn’t help but pry open his toys. It wasn’t about breaking them, it was about seeing how they worked. “I wanted to understand what was inside,” he recalls. That obsession never really stopped when I was a kid. It’s now more sophisticated.
Now, as a PhD candidate at Tecnológico de Monterey, Garza is pushing artificial intelligence to a whole new frontier: the microscopic world of chips, sensors, and embedded devices. Forget about large server clusters and data centers consuming megawatts of power. His vision is AI that can think locally, creating small, energy-efficient systems that learn and adapt on the fly, without relying on the cloud.
This bold idea earned him a 2025 Google PhD Fellowship. This prestigious award recognizes the most promising young scientists who are redefining computing over the next decade.
When Big AI gets too big
Most of today’s AI relies on huge computational infrastructure. This is like entrusting your brain to a giant digital “cathedral” – an endless rack of GPUs processing terabytes of data. It’s powerful, but it’s also unsustainable.
“Today’s large-scale AI models have a huge environmental impact,” Garza says. “We want to show that intelligence doesn’t have to mean excess. It’s possible to build systems that are just as capable, but much more sustainable and accessible.”
Introducing Tiny Reinforcement Learning (TinyRL), Garza’s minimalist twist on machine learning. Essentially, he’s teaching the microsystem to be smarter. TinyRL combines reinforcement learning (machines learn through trial and error) and Kolmogorov-Arnold theorem-inspired mathematics to enable embedded devices to optimize themselves in real-time. The most amazing thing about this process is that, unlike the large-scale machine learning systems that are popular today, it does not require a supercomputer.
A robot that learns by failing
In a university robotics lab, Garza and his team are testing small ground robots out of total ignorance. You don’t know where you are, how the wheels move, or what the sensors are for. But through thousands of small experiments of bumping into walls, rotating, and adjusting, you start to figure it out.
After a few hours of digital trial and error, that confusion turns into an adjustment. “You can literally see intelligence emerge from nothing,” Garza explains. Robots perform everything from nervous improvisation to purposeful navigation without the need for pre-programmed instructions or cloud-based training.
Soon, these algorithms will evolve to run on multi-microcontroller architectures, allowing multiple small agents to learn together and share discoveries, creating a kind of ecosystem of networked intelligence.
The human-centered future of Industry 5.0
This research underpins Tec de Monterrey’s Research Group for Industry 5.0, a collaborative effort to design technologies that are smaller, smarter and better for both people and the planet.
Garza envisions factories where robots learn new tasks on the job, homes where assistive devices adapt to users, and wearable health monitors that predict problems before they surface. “Imagine a smartwatch that does more than just track your pulse,” he says. “It predicts changes in your health and warns you before something happens.”
For Google, his selection as a 2025 Fellow makes him one of 255 doctoral candidates around the world working on pressing computing challenges. The program provides mentorship, funding and a global research network. For Garza Elizondo, it’s an affirmation that big thinking doesn’t have to be built into big machines.
“When people think of AI, they imagine a giant system behind a screen,” he says. “But what excites me is the idea that intelligence can exist anywhere, even in the tiniest corners of a chip.”
This article was written by Mexico News Daily staff editors in collaboration with Perplexity and was revised and fact-checked before publication.
