Embedded, Then Embedding
One word that followed me for decades — embedded system, then embedded Linux, then ML embeddings — each time a different creature. A look back — long before any of the tools we argue about now. For a few semesters in college I held a work-study job. Maybe it was because I was an out-of-town student who didn't live in the dorms; either way, the main task was carrying the mail that landed at the…
Embedded systems, embedded Linux, and embedding — three different incarnations of a single term that have evolved over decades. In my college years, I held a work-study job where my primary task was delivering mail and sorting it by lab. Among the odd jobs that came into the department was translating an article from a journal into Korean.
This marked my initial encounter with the concept of "embedded system." The article discussed performing high-level programming in C within the realm of sensors and controllers, small devices designed to handle more intricate tasks.
The main challenge I faced was translating the word "embedded" into Korean. At that time, the internet was not widely accessible, and the available English-Korean dictionary was outdated. After extensive research, I arrived at a Korean rendering that loosely translated to "built-in system." Interestingly, the concept of an "external system" had no direct counterpart in Korean.
Fast forward to around 2000, when embedded systems and embedded Linux gained significant traction. Companies began integrating CPUs and RAM to run Linux, and some even renamed their labs to incorporate the term "embedded." My own lab was once known as something other than "Computer Architecture and Networks," but the lab names changed frequently after I graduated.
During this period, my Korean rendering of "built-in system" gradually faded away. However, with the advent of smartphones and the shift towards more sophisticated products, the term lost its meaning in recent memory. As I delved into AI/ML work at a company, I encountered the term "embedding" again. Although sharing the same root as "embedded," embedding was a distinct concept rooted in pure mathematics and vectors. It involved reducing dimensions and did not necessitate any consideration of hardware.
In the era of large language models (LLMs) and ChatGPT, embedding has resurfaced. Although I initially hesitated when seeing "embed_model," people have since adopted the spelling "embedding." While some terms might be best expressed in English, others, like "embedding," have become a competitive advantage in the field.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
