The most important definition in the standard is that of written work. In this context, written work refers to selecting and arranging words in a meaningful order.
Language allows for an infinite number of word combinations in writing. When a human selects and arranges words, they are considered the essential author of that written work. Conversely, when a machine selects and arranges words, the machine is considered the essential author.
We support writers who use AI in non-standard ways and face challenges agreeing to the standard.
For instance, a poet may intentionally use AI to curate, compose, and recombine phrases to create innovative prose and poetry. They may find the VerifiedHuman™ label unhelpful on this work but use it on another.
In this way, we hope to be helpful and encouraging to all writers everywhere.
In this foreword, we aim to address two common cases.
i. AI Assistance in Drafting or Outlining Written Work:
It's a common practice for people to use AI assistance for outlining or drafting their writing. The VerifiedHuman™ Standard supports using AI for drafting and outlining written work. However, the final work presented as VerifiedHuman™ must be distinct and unique. We urge writers to cite their sources using appropriate usage and style guides, such as AP, APA, MLA, and CMOS. We recommend using the VerifiedHuman™ label with the AIA (AI-Assisted) tag: VerifiedHuman™AIA. However, it's at the writer's discretion when the tag is helpful.
ii. Using Content of Unknown Origin in Written Work:
Sometimes, writers use phrases considered common knowledge source material of unknown origin. In such cases, the writer is unsure where the words came from. In other situations, the writer may suspect that AI has generated some of the material they are using, but the origin, time, place, purpose, etc., are unknown or uncertain. We suggest writers use the VerifiedHuman™ label with an AIU (AI-Unknown) tag: VerifiedHuman™AIU. But, again, it's at the writer's discretion when the tag is helpful.
We believe the intention to create and present written work as their own should always remain with the writer. Therefore, we urge all writers to follow the Five-Word Principle and iterate their work in the writing process to ensure that their final product is unique and distinctive.
Here are commonly accepted ways AI assists in the writing process.
Drafting and outlining can now be done using AI-powered tools. These tools can help frame reasons or arguments or outline and organize multiple ideas.
Grammar assistance is another area where machine learning can be useful. AI-powered tools can suggest better or different ways of saying what a person has written.
Interpretation refers to the process of understanding or explaining the meaning of something.
Language translation involves translating words and phrases from one language to another.
COMMON USES OF AI IN WRITING
Here are two questions that the AI-using writer has to interpret independently, and the author is responsible for following their values and judgment.
1. Can I use ChatGPT to write an outline or paper and then rework it in my own words?
Some people use language modeling tools to help them organize their thoughts and structure their writing, especially for lengthy, complicated, or unfamiliar work. They ask the AI to write a draft and then personalize or contextualize it to some extent.
2. Can I have ChatGPT write a paper, read it, and then write my version?
Some people ask language modeling tools to write a draft of what they want to say and then use it as a guide or model for organizing or approaching their work.
While using these approaches is becoming more common, caution is recommended. The danger in both cases is that it can become difficult to distinguish between a person's thinking and the AI's "thinking."
If you’re copying and pasting entire thoughts or sentences, you’re not doing the writing. Instead, you’re copying writing done by someone or something else. As a general rule, we would suggest that if AI writes five (5) specific words, selects the words, and puts them together into a phrase, then AI wrote the phrase. Five isn't a magical number. It's just enough to be a considerable phrase or thought. The spirit of the principle remains in the question, "Did you write it, or did AI write it?" no matter how many words are in play.
THE FIVE-WORD PRINCIPLE
In writing, authors put words together to write lists, lines, sentences, and paragraphs or put sections together to form longer works like letters, papers, reports, and books. Essential authorship of any written work can be established by asking a simple question.
The essential question of authorship is: Who put the words together? A simpler way of asking is: Who wrote it? This applies to entire written works or even small portions of written work.
The basic unit of language is the word.
Words are put together to form ideas. These ideas can be complete sentences or just collections of words in a line (like phrases).
When lines of words or sentences are arranged meaningfully, the result is some form of written work.
ASSUMPTION OF HUMAN AUTHORSHIP
If a human puts words together into lines, phrases, or sentences and arranges them meaningfully, then a human is the essential author.
ESSENTIAL HUMAN AUTHORSHIP
Other definitions of words and ideas related to the Standard for Writers are here.
Artificial Intelligence (AI)
A field that combines computer science with robust datasets to enable problem-solving. It includes sub-fields like machine learning and deep learning, frequently mentioned in conjunction with AI. These disciplines comprise AI algorithms that seek to create expert systems capable of making predictions or classifications based on input data.*
AI language modeling (or Large Language Modeling (LLM))
Refers to systems that generate natural language texts from large amounts of data. Large language models use deep neural networks, such as transformers, to learn from billions or trillions of words. They can produce texts on any topic or domain and perform various natural language tasks, such as classification, summarization, translation, generation, and dialogue. Some examples of large language models are GPT-3, BERT, XLNet, and EleutherAI.**
A personal, original idea
Represents a specific human being's unique insight or experience in the world.
Another area where AI-powered tools can be useful. They can correct misspelled or mistyped words.
Principles or standards of behavior.
Principles, values, or ethical assumptions that motivate human behavior.
A single distinct, meaningful element of speech or writing.
*IBM: What is artificial intelligence (AI)?
**Microsoft | Learn | LLM
This document provides clarity on terms in the standard and their meanings. It also helps writers understand the motivation behind the standard.
What's in this document?
Here are definitions of words and ideas used in the Standard for Writers.
A specific definition of human behavior
A person who authors written work
Present, share, or show work with others
The selection and arrangement of words in a meaningful order. Lines and phrases that have been written down, typed, or transposed by putting words together
A group of people working together
A work or invention that is the result of creativity, such as a manuscript or a design, to which one has rights and for which one may apply for a patent, copyright, trademark, etc.***
The fundamental elements or characteristics of something
Selecting or arranging the essential component of writing–words–into a meaningful order to create written work (see Essential Authorship below)
Essential elements of writing
Noun: a human person; adjective: from a human person
Shorter definition–machines that create novel (new) content | Longer definition–generative artificial intelligence (AI) describes algorithms (such as ChatGPT) that can be used to create new content, including audio, code, images, text, simulations, and videos.****
"Systems that act like humans"*
Other generative processes
Other processes involving AI or machine learning to create novel (new) content
*IBM: What is artificial intelligence (AI)?
****McKinsey & Co. | What is generative AI?
DEFINITIONS IN THE STANDARD