In a small room at the National Polytechnic Institute of Cambodia (NPIC), a box-like robot bearing the Cambodian national flag and logo of the institute on its 3.2-inch screen showcases its remarkable capabilities.
Created by a team of NPIC students, the robot can understand and respond in Khmer, a testament to the strides Cambodia is making in both artificial intelligence (AI) and robotics.
During a demonstration, a student greets the robot with “chhum reap sour” (hello), and the robot promptly replies with "hello”.
When the student says “chhum reap lear” (goodbye), the robot responds with "learheoy" (bye). Upon hearing the word “Cambodia,” the robot expresses its affection, saying "I love Cambodia."
Ny Virbora, an electronics engineering professor at NPIC who led the four-member student team that created the robot, explained their objective.
“We wanted to create a Khmer automatic speech recognition system that can turn spoken Khmer into a written Khmer transcript generated by itself without using the internet,” he said.
Leveraging their studies in software programming and AI, they embarked on the innovative venture.
Bora elaborated on the technology behind their creation. The AI used in this project functions as an internal operating system, designed to operate independently within the machine itself.
Two key components were integrated to achieve this: Khmer Automatic Speech Recognition and Khmer Text to Speech systems.
“The Khmer Automatic Speech Recognition is an AI system that trains voice data types for each word, and once training is complete, the AI can accept audio input and output characters matching the spoken words,” said Bora.
“The Khmer Text to Speech component is crucial for enabling the robot to respond audibly, as it converts characters to sound, allowing the robot to vocalise responses based on the input from the speech recognition system,” he told The Post.
Specialised coding education
Leng Piseth, co-founder of the SabaiCode IT education initiative, shared insights into their AI-related projects, saying that AI is a huge focus at SabaiCode.
“We view it as an amazing technology revolution and want to leverage it to serve our students better,” he told The Post.
“We are currently developing an advanced e-learning application similar to a learning management system, which will either develop or integrate AI systems designed to enhance the learning experience,” he added.
These AI systems will offer personalised learning paths, adaptive content, and intelligent feedback mechanisms, ensuring that students can learn more effectively and efficiently.
At SabaiCode, they recognise the potential of AI and aim to build a solid foundation for their students.
Two years ago, they launched a course titled "Foundation of Data Science," where students begin with Python programming, basic statistics and data visualisation.
Additionally, they offer a data engineering bootcamp, which was successfully delivered to engineers at EZECOM.
The trainees learned to create data pipelines and develop various machine-learning projects, showcasing practical applications of their AI knowledge.
A university perspective
Chan Sophal, a lecturer of data science at the state-run Institute of Technology of Cambodia (ITC), presented an insightful overview of AI and its applications to his students.
Sophal began by defining AI as the simulation of human intelligence processes by machines, particularly computer systems.
He emphasised the need for specialised hardware and software to develop and train machine learning algorithms, noting that popular programming languages such as Python, R, and Java are used in AI development.
Highlighting the evolution of AI, Sophal said, "In 2023, we saw significant strides with the emergence of generative AI technologies like ChatGPT, Midjourney, and DALL-E. These advancements have opened new possibilities for AI applications across different sectors."
Sophal's presentation delved into practical applications of AI, showcasing its impact on various industries.
In the healthcare sector, AI is revolutionising patient care through advanced diagnostic tools and personalised treatment plans, while the entertainment industry is leveraging AI to create more immersive and interactive experiences for users.
Additionally, AI is transforming agriculture by enabling precision farming techniques, which enhance crop yields and reduce resource consumption.
He also included demonstrations of AI in action.
“We featured facial recognition technology and self-driving cars. These examples illustrated the practical benefits and potential challenges of integrating AI into everyday life,” explained Sophal.
Addressing the ethical considerations of AI, Sophal underscored the importance of developing AI systems responsibly.
He advocated for ethical guidelines to ensure that AI technologies are used for the greater good, emphasising transparency, accountability and the mitigation of biases in AI algorithms.
AI in the classroom
AI has revolutionised the way students improve their reading abilities with the introduction of "Akhra," the first mobile technology education app developed by the Centre for Digital and Distance Learning of the Ministry of Education, Youth, and Sports.
Akhra utilizes AI to support reading practice through various features tailored to enhance learning outcomes.
The app provides text at different difficulty levels from the Khmer textbook, allowing students to practice at their own pace.
It includes word recognition and explanations of difficult words, ensuring that students understand the text thoroughly.
AI-generated sample reading sounds help students improve their pronunciation and listening skills.
The app offers quiz practice aligned with international reading proficiency tests, enabling students to gauge their progress.
For essay writing, students can submit handwritten essays and receive direct feedback from teachers.
They can also practice reading aloud and send voice recordings to teachers for further improvement and evaluation.
AI technology in Akhra allows students to measure their reading accuracy and speed, providing a comprehensive score that reflects their progress.
Kingdom's AI landscape: Applied user, not developer
Bora explained that the initial AI algorithm for the robot was developed at Carnegie Mellon University in the US and served as a foundational software package for training AI in automatic speech recognition.
However, the NPIC team significantly enhanced this AI by combining two major algorithms and methods they developed: Dataset and data management software for data collection and text to speech for voice responses.
Through local enhancements and customisations, the NPIC team successfully created an AI system capable of facilitating voice recognition and response in Khmer, a significant step.
However, Cambodia is not yet a developing hub for artificial intelligence (AI) but rather an applied user.
Srun Channareth, a professor at the National Polytechnic Institute of Cambodia and the Department of Electronics Master of Engineering, provided a candid assessment of the AI landscape in Cambodia.
“We do not have our own AI tools; most of them are developed by international companies,” he told The Post.
“We use Applied AI, which means they are formatted, and we just apply them,” he added.
Lim Sangva, co-founder of innovation centre ArrowDot, provided a realistic view of AI's current usage in Cambodia.
He explained that the Cambodian market does not yet utilise key AI features. Its usage is currently at the Robotic Process Automation level, which cannot be considered AI at a system level that analyses data, as it follows a pre-designed system dependent on user settings.
“We still limit the use of coding or programming; it is not yet at the AI level. What we use daily and teach is not yet integrated into AI technology; it is at the level of Robotic Process Automation,” he said.
He added that neither software development nor hardware has integrated AI technology, because it does not analyse data or make decisions based on it.
“Currently, ArrowDot uses some AI functions to automate data analysis and related business development work. I use AI functions to automate some content for marketing materials and proposals,” he added.
Future prospects of AI in Cambodia
SabaiCode’s Piseth believes that Cambodia is just beginning its journey to develop talent and leverage AI technology.
“The country needs to build a robust data infrastructure to efficiently collect, clean and store data, which is crucial for training and developing AI systems,” he explained.
He said the growing interest in AI within Cambodia is very encouraging, and there is optimism about the potential for significant advancements in this field.
Sangva noted that his team is still researching the technology needed to keep pace with global developments.
The development and implementation of AI technology in Cambodia are still in their nascent stages, but the efforts by institutions like NPIC, SabaiCode, and ArrowDot highlight a promising future.
With dedicated educators, innovative startups and supportive government initiatives, Cambodia is slowly but surely positioning itself to harness the power of AI.
As the country continues to build its data infrastructure and train its young talent, the revolution of AI in Cambodia promises to bring about significant advancements across various sectors, driving the nation forward into a new era of technological innovation.