When AI needs more than intelligence—it must understand human diversity.
In today’s world, AI has become part of daily life—from recommendation systems on various platforms to intelligent assistants that can talk to us. But a crucial question remains: do these AI systems truly understand us?
Especially when “us” doesn’t mean a single, uniform group—but rather a wide range of people with different languages, physical abilities, beliefs, and identities. This is where the concept of Inclusive AI begins—artificial intelligence that leaves no one behind.
What is Inclusive AI?
Inclusive AI refers to artificial intelligence systems that are built with an understanding of human diversity—whether in terms of physical ability, language, culture, gender identity, or social context. It’s not just a matter of technical performance, but a process that begins by asking: “Who gets to access this system?” and “How well does this system understand people who don't fit the norm?”
Inclusive AI must be accessible—free from technological or interface barriers that may be too complex for certain groups of users, especially those who are often overlooked by conventional systems, such as persons with disabilities or people who do not use formal language. It must also be free of bias embedded in legacy datasets, which often arise from training models on limited and non-representative samples of society.
Most importantly, inclusive AI must be developed within a clear ethical framework. It's not just about building the fastest or most accurate system, but ensuring respect for user rights, transparency in data collection and use, and fair value sharing with the communities that contribute to these systems.
How Do We Build Inclusive AI in the Thai Context?
Thailand has its own unique challenges when it comes to AI:
- • The Thai language includes multiple levels of formality, regional dialects, and varied expressions.
- • Many people with disabilities still face barriers to accessing technology—or are viewed as a “special case” that the system doesn’t need to accommodate.
- • หลายระบบยังออกแบบบนพื้นฐานของ “คนทั่วไป” โดยไม่ตั้งคำถามว่าใครคือคนทั่วไปกันแน่
Building Inclusive AI in this context requires more than just technology—it requires a new approach.
How Is Vulcan Building Inclusive AI?
Over the past six months, the Science team has been developing and testing a Chat Intelligence system with a clear goal: To reduce errors in message routing and increase service speed. The focus is on helping the system understand questions that don’t follow conventional patterns, such as:
- • Users who ask in incomplete sentences, like “Shipping fee?” instead of “What is the shipping cost?”
- • Messages with typos or rushed phrasing—the system must still understand and respond accurately.
Challenges in Building AI That Truly Understands Thai Users
1. Thai is a complex language for AI. Thai has no spaces between words, no clear prefixes or suffixes, and many regional accents. The team must test various methods to accurately interpret intent and classify messages.
2. Not all users are the same. Some users speak briefly. Some use slang or niche terms. Others—such as people with certain disabilities—may struggle to type. The system must avoid crashing or failing to understand simply because of these differences.
3. AI that understands = AI trained with real user input. The team adopts a “test first, scale later” philosophy, prioritizing real-world trials over theoretical perfection.
From Chat Systems → to Human-Centered Systems
The Science team’s goal for the next quarter is simple: "Get things done." But beneath that simplicity is a deeper mission—to deliver systems that empower service teams, help users access information more efficiently, and reduce repetitive problem-solving.
เพราะ AI ที่เข้าใจความซับซ้อนของมนุษย์ ไม่ใช่เรื่องหรูหรา แต่มันคือพื้นฐานของระบบที่เข้าถึงได้จริง หากคุณอยู่ในสายงาน AI, UX หรือ Operation อย่าลืมว่า “Inclusive AI” ไม่ได้เริ่มจากโมเดลที่ซับซ้อน แต่มาจากการฟังคนที่ระบบมักมองข้ามและนั่นคือสิ่งที่ทีม Science ของ Vulcan ทำอยู่ทุกวันอย่างเงียบ ๆ แต่ไม่ธรรมดา
