Intel Brings Offline AI, Opportunity to Students in Guatemala
An initiative using Intel Core Ultra-powered AI PCs bridges the digital divide through on-device AI accelerators, bringing advanced technology to remote locations.
via Intel
AI is reaching students in Guatemala, without internet.
In and around Guatemala City, a vibrant hub of arts, music, food and family, a stark reality persists: 57% of its 3.2 million metro-area residents live in poverty, many of them children. They are among the 1.3 billion children globally who lack internet access – and the gateway to education and opportunity it offers.
But in one corner of the city, children benefit from a new initiative that's bringing the power of artificial intelligence directly to students without an internet connection. Students at Esperanza Juvenil, a private nonprofit school and residential program, now learn with AI, thanks to a donation of Dell laptops powered by Intel® Core™ Ultra processors.
These aren't just laptops. They're AI PCs, equipped with a central processing unit (CPU), graphics processing unit (GPU), and a neural processing unit (NPU) to run AI tasks directly on the device. Each is loaded with an offline large language model (LLM) tutor developed by World Wide Technology, offering personalized, portable learning designed for places where connectivity can't be counted on.
The solution is designed to be scalable and affordable, allowing for expansion to new schools, subjects, countries and languages. Open source solutions are prioritized to avoid vendor lock-in and minimize costs.
Empowering Education Through AI
"These are children who learn to overcome many challenges, who are eager to change their story and their family history through quality education," says Susana Rios, director general of Esperanza Juvenil.
Lucas Pinzon, the school's executive director, adds, "We firmly believe that a good education can open doors for them."
The specially trained AI tutor program is already making a difference. The LLM helps students improve their Spanish, learn English and math, and explore the world around them. Teachers can analyze the chat logs to gain insights into students' interests and performance, allowing for more personalized lesson plans. And the AI tutor is designed to understand Guatemala's cultural differences.
"Imagine a more personalized, nuanced AI that has been made for Guatemala, that's been tuned and adapted, that knows the cultural variances of Guatemala in order to be able to better answer some of the regional questions that these students may have," says Bharath Srivats, an AI product marketing engineer with Intel. "All of this power is now portable and can be brought on device – AI on the go."
Srivats, who grew up in a rural town in southern India facing similar educational resource limitations, deeply connects with these students' experiences. "Had I had internet back then, I could have learned some things. But if that kid had an AI back then, that kid's life would have radically changed. That kid would've had answers to everything."
The Future of Offline AI
Bob Olwig, executive vice president of World Wide Technology, one of Intel's largest data center partners and the LLM's designer, acknowledges the logistical challenges for an application that needs to run flawlessly without tech support in remote locations.
"There are a lot of obstacles to getting this done," he says. "Our goal really is to democratize technology so that students and school kids anywhere around the world can have access to leading-edge artificial intelligence applications. We're really proving out architecturally how Intel's technology can support running LLMs at the edge, which is super exciting."
Beyond academic tutoring, the potential for offline edge AI extends to various fields, including remote research projects, manufacturing plants, mining, tunnel work, search and rescue missions and rural farming.
Srivats highlights a key benefit of this Intel AI solution: the total cost of ownership. "By running more of your workloads on the device, relative to cloud, for the sunk cost of the device, you are now running AI for free with these free open source models," he says.
This initiative in Guatemala underscores how on-device AI accelerators can power AI-enabled features at the edge, making advanced technology accessible where it's needed most. Intel and World Wide Technology are set to expand these donations to schools in India, Africa and South America.