IRTNOG and LLMs E014

Просмотров: 5   |   Загружено: 10 мес.
icon
A Bit of Security, by William J. Malik
icon
0
icon
Скачать
iconПодробнее о видео
IRTNOG and LLMs
A Bit of Security for April 2, 2024
E B White’s short story IRTNOG was published in the November 30, 1935 New Yorker magazine. It tells us what the end game is from LLMs and reductive models of AI. He wrote Charlotte’s Web, and co-authored the wonderful book “Elements of Style.” In the short story, he looks at the world of the 1920s, when the volume of information is so great that no person can keep up, someone develops a machine that can digest an article and publish the brief synopsis. Read the synopsis, and you know all the information in the article. Unfortunately, the amount of information continues to increase, so another machine comes along that digests the synopsis into an even further concentrated gloss, and eventually we end up with a master machine that can take all the printed information every day and produce a single word that encapsulated all events of the past 24 hours. They turn the machine on, and it writes: IRTNOG.
This reminds one of the central conceits behind The Hitchhiker’s to the Galaxy. Deep Thought answers the question “What is the meaning of life – the Universe – and everything? With the enigmatic “42”. Now, humanity has to build an even more powerful computer to explain what that answer means.
ChatGPT and LLMs generally do nothing more than digest and condense information until they leave us with a nonsensical result. The collapse of LLMs follows from the pollution of their source material, which is already so polluted by the output from other LLMs that the results are three or four standard deviations away from authentic human constructive thought. Ludwig Wittgenstein suggested that language is a game we play collectively. Playing against a machine is ultimately sterile and fruitless. Meaning resides in intentionality, not statistical inference.
As an aside that’s why speaking in tongues is less important that interpreting. As St Paul wrote in I Cor 14:9 “Unless you speak intelligible words with your tongue, how will anyone know what you are saying? You will just be speaking into the air.” The Philosopher Martin Heidegger lectured about the essence of meaning in a set of essays collected under the title “What is Called Thinking?” There, he quoted Parmenides, in a fragment which may mean “To think of something it must exist” – LLMs, buried in silicon, cannot conceptualize the ideas of nothingness, of non-existence, or of contradiction. Instead, they rerun statistical autocomplete eternally, eventually distilling all texts into progressively less meaningful utterances.
For now, the primary dictum governing unbounded LLMs seems to be the Talking Heads: Stop Making Sense! IRTNOG!
Let me know what you think in the comments below or at wjmalik@noc.social
#LLM #ChatGPT #AI #BitofSec #cybersecuritytips

Похожие видео

Добавлено: 55 год.
Добавил:
  © 2019-2021
  IRTNOG and LLMs E014 - RusLar.Me