Making Up Words or Inventing a Vocabulary
Or why Large Language Models are useless
All words in all languages are invented. They are all invented by humans. Different nations (or cultures) have different words, invented by different people. The same (identifiable) things have many words that refer to them, sometimes several words in the same language but always several words in many languages. Sometimes the words resemble an objective, audible sound, e.g. "hiss".
There is nothing "true" or "objective" about the words themselves. Many people wrote many dictionaries over the years. Several people even invented words like "dictionary" or "word". It's a bit recursive in a way. At their core, these things are arbitrary, subjective, even self-serving.
Many people invent words, but those people cannot persuade enough people to adopt and popularise these words. So those words just "die" with their inventor (or few adopters).
High-profile people (or companies) can make words stick. For instance, Richard Stallman came up with many words and acronyms. Some are not proper words but jokes or wordplay. He has many Web pages that try to define and spread those.
All language die one day. They are more or less officially dead when nobody speaks them anymore or when books documenting those languages (e.g. dictionary) become unavailable or read by nobody.
Programming languages are not languages per se, even if they resembles languages. They're instructions that are convertible to low-level code organised for serial processing on some machine (and different machines use different low-level code).
Concepts matter more than words.
Ideas or methods transcend languages.
Society should focus more on equipping people with skills, not training them to memorise acronyms with which to communicate brands and superficial industrial/societal constructs.
Education is about enhancing the ability of person, not teaching the person how to use tools like chatbots, which aren't mnemonics but mere "BS generators" as Dr. Richard calls them. He can speak quite a few languages, but what makes him stand out is his careful attention to details and meticulous criticism of Bad Ideas. Being pedantic makes him effective.
Large Language Models or chatbots (the latter is one use case of the former) lack comprehension of languages and concepts. Just because they can string/cluster together a bunch of words and adopt ridiculous buzzwords such as "super intelligence" doesn't change the fact that they "master" what matters far less: words.
Concepts, not words, are what "makes perfect". Excellence through plagiarism is cheating. Emulation by plagiarism is for scoundrels. █