Conserve usage of large-language models
![]() |
| Large-language models return answers based on their databases and protocols |
A large-language model (LLM) is a software system that generates an impression of conversations using petabytes of data stored in its databases. The concept of LLM emerged in the early 2020 with the advent of OpenAI ChatGPT, which gained 100 million users in its first month of operation. Having noticed the advent of ChatGPT, several software engineering companies also launched their own large-language models. Large-language models rapidly generate answers but their underlying infrastructure consumes billions of watts because they are thousands of servers that must transmit data across tremendous distances and cable lengths. As for their massive energy consumption, their companies commissioned power plants specifically to serve the power requirement of their LLM projects.
LLMs are best for generating article summaries or rendering information for relatively non-divisive topics, such as how to stay healthy or how to display source code. There are also great for rewriting information that do not require intensive human consideration, such as mathematical formulas or calculations of numbers. As LLMs are a recent invention, it is safe to assume that they are unreliable for "evaluating" the minute details or fringe parts of human knowledge at large. LLMs cannot "know" or "judge" anything; they only follow protocols that their developers insert into their server-side software.
Governance
Do not use large-language models to generate conclusions or judgments about human governing bodies. If you prompt about a nonstandard governing body or political organization, LLMs behave like state loyalists. The answers they generate are out of context at best, despicably offensive at worst, as their databases and protocols force them to ignore the context of nonstandard poltics and instead judge via the government/state framework. Governance is a purely human endeavor. Except for governments and sovereign states, no LLM can give you accurate and context-aware interpretations about the governing bodies or political organizations that you present to their prompts. Instead, consult people on how to rule, govern, or manage communities. Official statements or issuances by governing bodies or political organizations are safest when edited with human creativity. If you are called to draft an issuance for a governing body or a political organization, you have to edit manually, like LLMs were not yet a thing.
Health
| If you have serious health issues, immediately seek attention from a licensed physician or health authority |
|
Department of Health https://doh.gov.ph/ |
| Research Institute of Tropical Medicine https://ritm.gov.ph/ |
LLMs are for informational purposes only. They can generate information about health conditions, but they are not designed to be consulted. You can upload information about symptons and causes of disease into an LLM prompt, but do not use LLM outputs to decide for your health. Instead, stick with obedience to your physician or health authority. Humans know better than chatbots in matters of health. Everyone has their own health requirements that LLMs cannot comprehend at all. If you suspect you have a health issue, you should visit a physician for a diagnosis. On substance consumption, you can take food supplements or drugs for as long as your physician and health authority expressly allow that. Before consuming any food supplements or drugs, always tell your physician about those substances; the physician reviews the nature of those substances before releasing their verdict. When the physician says no, that means no. Just do not consume the vetoed food supplement or drug. Silence or the absence of a response also means no. Remember, food supplements and drugs interact with each other.
Arts and design
This is where the stakes go sharply higher. Using LLMs to render drawings consumes tremendous amounts of energy and is a catastrophic risk to human artists. If you have a decent income source and you want to own a visual artwork or a piece of music, you should consider commissioning a human painter, sculptor, composer, or some other kind of artist to craft for you. If you want a building design that accurately accommodates your requirements, there are architects waiting for customers and you should contact one of them. If you own something like an accommodation business, you can hire a band, a choir, or orchestra to play music for your guests. Nonstandard governing bodies? Perhaps it's time to give your state minister or legislator another more creative and potentially more noble purpose: design novel unique governance or political systems from scratch. By the phrase "from scratch," that means truly novel frameworks, not a spinoff of the government/state framework. Also remember what I said earlier: do not use LLMs to evaluate nonstandard politcs.

Comments
Post a Comment