In any domain of knowledge, we use many words and terms to describe the concepts, techniques, etc., related to that field. Afaik, terminology deals with words to be used in a context and their corresponding meaning.

Etymology is fundamentally different from Terminology. In general, terminology doesn't address the origin of the word in a particular context.

Some words carry information related to the phenomenon under consideration. For suppose the term gradient descent contains the gist of the algorithm. There may be many terms that are unrelated and may be of historical origins, such as the name of a scientist or some other thing.

If a user is confused about the words or terms or even phrases that are used in AI, and asks about the origin or the intention of selecting a particular term, then is it on-topic?

Example questions:

Why do we use the word gram in n-gram?

Why logistic regression is called so?


2 Answers 2


I am in favour of this type of question being on-topic here. In particular, it seems that those questions would be at the intersection of "history of AI" (which is on-topic here) and "terminology" (in the sense that you want to know something about terms). We have had some of those question in the past. I remember one that I actually provided an answer to which was never closed as off-topic and I had never considered it as such: Why is it called back-propagation?.

  • $\begingroup$ By the way, I am currently reading a book about the history of AI and, occasionally, the author provides info about the origin of a term. $\endgroup$
    – nbro
    Commented Aug 9, 2021 at 0:11

Absolutely! This is an area where I have formal training, and can legitimate go back to proto-indo-european. (But people typically don't like my analyses b/c they conflict with their own.)

If you see any unanswered terminological question, or are interested in the etymology, this is an area of special focus for me in the tech industry and academia, as is the history of computing.

Clear terminology is critical for any technical field, and good terminology is supported by etymology.

One of my favorite topics is what "Artificial" and "Intelligence" mean, and what they mean together. My research indicated the latter goes back to the pie *legh/-leg as "discrimination in the sense of selection". Think gatherers choosing which berries to gather from a set of berries, or hunters identifying the most vulnerable member in a herd. (These are combinatorial problems in a naive sense, and definitely game theory problems—maximizing the minimum reward, or minimizing-the-maximum loss.) The Latin prefix "inter" means "between" or "among" in the sense of "matters", originally legal, lexical and so forth, which is to say "intellectual", because the Latin lex comes from the Greek logos. I even argue that, as a function, it is a grounded symbol. I express it M(u): "in an action space, utility", which can then be extended as with Hutter&Legg's formal definition of universal machine intelligence.

I see other interesting etymologies. As a student of Norse mythology, I can't help but think about the Norns (Fates) when I see the word "entanglement". Heisenberg pretty much coined it, if I recall correctly, and, because Old Norse, Standard German, and English all come from the Germanic branch, it translates quite precisely. I don't see this as a random choice, because H was talking about probabilities, which is a commentary on fate. The Norns wove (entangled) and cut (limit) the fates of human beings. I feel like Norse mythology allowed for free will on a local level (choosing to die in battle to gain entrance to Valhalla), as opposed to a global level (Ragnarok is predetermined.)


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .