Was AI Included?

It’s a simple question and one that I think we all as artists, writers, knowledge workers and, frankly, all tech-enabled humans need to start answering.

Much like food companies are required to assure food is safe and disclose the main ingredients and nutritional stats on what we eat, it’s time for all of us to start acknowledging and disclosing when and how we use AI in the work we produce.

AI and other generative technologies present us as a society and as creatives with a lot of hard questions, spanning authenticity, intention and whether or not to disclose when and how we use these AI tools. We need to start by asking questions to define how we want a world of AI creations to work. As it relates to disclosures, we might start with:

• When we create something, should we be sharing whether AI was included? Or should we just stealthily act as if we did it all? Or perhaps something else?
• If AI did have a role in your creation, how to acknowledge its role? As a contributor, researcher, collaborator or something else?
• If you are a publisher of art, information or content, should our audience, consumers, followers and the general publish be informed about AI being included? Why or why not?

In this “Age of AI,” it is becoming increasingly fuzzy to know if a piece of art, writing, or even a chat message was created by a human, by AI alone or a hybrid of both. As creatives, it’s equally confusing to know where we should stand as it relates to AI and other creative technologies too, both authentically and practically. Is AI taking our jobs or making our jobs easier and better? Is AI a tool of human empowerment or the opening salvo of AI overtaking humans?

We need a new social norm of AI usage for both consumers and creators of art and information.

There is a lot of tension and ambiguity personally and professionally using AI and consuming AI creation. Recent AI tools have dramatically changed what it means to and feels like to “create,” resulting in considerable confusion, misinformation and uncertainty around AI, art, writings and even self-expression.

In this post, I want to argue in favor of a new social norm for creatives and knowledge workers. Specifically, I believe we should start aknowledging the use of AI tool in our creative output and productive work as well as, to a limited extent, the role of AI and tedchnology in our creative processes. I’ll take the position that there should be greater transparency for when AI gets used and how, and the form of disclosure I recommend and use I’ll call “AIDA.” It stands for “AI Included Disclosure Acknowledgment” and is a tribute to Ada Lovelace, a 19th Century female computer pioneer.

Let’s start with some definitions and some historical context building.

Social Norms, informal rules of behavior shared by a community

What is a social norm?

Social norms are unwritten rules of behavior that are shared by a group, community, or society. They are largely informal, shared understandings that guide human behavior by defining what is acceptable and appropriate. For example, being on time, offering your seat to the elderly or disabled, or even certain fashions.

Sometimes social norms get codified into rules and laws in the form of a professional code of ethics. Phenomenologically, social norms offer us a socio-ethical framework for intersubjective meaningful action. They form the bedrock of community too.

Social norms evolve over time, often in response to changes in society, nature, economy, and technology. A social norm exists when individuals practice a behaviour because they believe that others like them will follow that same behavior. They are meant to align us to live and work together in a harmonious and just community.

Defining and following new social norms offer us a way forward when past best practices no longer serve, new technologies shift the landscape, and we need to re-orient our actions individually and collectively.

Attribution, Acknowledge and Citations: Definitions

How does one reference and give credit to our influences and originators of certain ideas and cited words in one’s academic, scholarly, or creative works?

Attribution, acknowledgment, and citation are all varying forms of giving credit, disclosing and/or referencing to others for their influence on what we create. They have a long history and evolution. There are three primary manifestations and words in English:

1. Attribution is the broadest term and refers to the act of recognizing and identifying the source or origin of a particular idea, concept, work, or contribution. It involves acknowledging the source or creator responsible for the content you are using or referencing.
2. Acknowledgment is a specific type of attribution and often occurs in the context of expressing gratitude or recognizing the contributions of individuals or organizations. For example, those who have provided support, assistance, or resources for a project, research, or creative endeavor.
3. Citations are a formal and systematic attribution method for giving credit to the sources of information, ideas, data, or quotations used in academic or scholarly works. There are specific citation styles (e.g., APA, MLA, Chicago) that are intended to make it easy for readers to locate the original information.

Each of these types come with pro’s and con’s. At their core, attribution practices offer us ways of respecting the contribution of others and openly recognizing it.

Now that we’ve defined some key terms, where did the idea of such disclosures come from and how has it evolved?

A Brief Note on the History of Attribution and Disclosures

The history of acknowledgment and attribution practices is closely tied to the evolution of human communication. It intersects with history and technologies of knowledge dissemination (manuscripts, printing, computers and the internet) as well the development of societal and professional social and ethical norms.

In ancient civilizations, most communication was verbal. During the oral telling of a story, sharing an idea or a teaching, the storyteller traditionally acknowledged the source or authorship. This was part of cultural practices and rituals and was meant to convey respect to our ancestors, recognize past innovators and empower leaders. It was a recognition of the gift of a story, idea or innovation, and it glued together a society.

With the invention of writing and rise of philosophy and science in ancient Greece and Rome, scholars began to attribute certain ideas and writings to specific individuals. For example, Plato in his early dialogues would include Socrates and others as dialogical characters. He would name them and give them a voice in the philosophical debate as a way to indicate the source of a philosphical position.

Over time, citations of source and quoting became more formalized. This became more and more apparent in scholarly works with the goal of establishing authority and recognizing credibility of past thinkers.

During the Middle Ages, manuscripts and books were painstakingly copied by hand by scholars. During the process, scholars would often include colophons (notes at the end of a manuscript) to attribute the original author of a work. As theological traditions become more sophisticated and debates would traverse different locations and ages, proper attribution was further enshrined culturally. It was embedded in the Medieval scholarly tradition. It become an expected practice or social norm to cite and acknowledge the originator of an idea, author of book, etc.

During the Renaissance, the invention of the printing press in the 15th century revolutionized the dissemination of knowledge. Printed books became more accessible, and authors and publishers began to assert ownership over their works. Specifically, title pages and prefaces were used to attribute authorship, recognition of the publisher, date and edition.

Furthermore, in the 18th and 19th centuries, patent and copyright laws were established which granted legal protection to inventors and their inventions and to publishers authors and their works. Essentially this established the right to control how their creations were used and attributed and further shaped acknowledgment practices. It also established the idea of “owning” an idea or creative output.

As scientific societies and academic publications were formed in 19th and 20th centuries, more formalized attribution practices in scholarly research developed too. It became standard practice in academic writing to give credit to previous studies and acknowledge the foundation of knowledge upon which new research was built using citations and references in the text or at the end. For writers, artists and scientists, these practices have evolved over time to address issues of intellectual property, authority, and fairness in the sharing of ideas and information.

Computers, the internet and the digital age put new pressure on acknowledgment practices as well as led to new opportunities and approaches. It become easier to share content, ideas and media online through forums, social media, websites and more. Online encyclopedias and websites made knowledge and information more available and accessible. This also led to plagiarism and unauthorized use too. Ideas, words and content spread without payment or acknowledgement and often misattributed.

Along with rise of open source software, the open-access movement and Creative Commons licenses have introduced new models of sharing and attribution in the digital realm. These frameworks allow creators to define how their works can be used and require users to provide appropriate attribution when that work does get used. For example, much of my own writing and coding projects are explicitly shared so others can reuse and remix them freely as long as they provide some acknowledgement.

Throughout history, acknowledgment and attribution practices have evolved to reflect changes in communication technologies, cultural norms, legal frameworks, and ethical considerations. Disclosures and acknowledgment are not only a matter of intellectual property but also a way to honor the contributions of others and maintain the integrity of knowledge dissemination. It’s also a way to recognize the role of past thinkers and creatives have on our work and thinking. The core of these practices create a social norm of expected behavior in the writing and sharing of content, art and ideas.

This leads us to AI. Much like these past attribution practices were tied into evolving technologies, like printing presses, legal changes, and new social norms for scientists and scholars, Generative AI text and art represent a major societal moment. It feels like our past social norms are insufficient for AI creations and present hard questions on authenticity and AI usage too.

What are some options for navigating the ambiguity of attribution in the age of AI?

Why Disclosures are Needed and Necessary: A viable path forward for creatives using AI tools to create

There are now well-established social norms for citing a quote or attribution in academia and the sciences, but what about AI attribution? Is it necessary? And what are some options?

Now that we’ve considered some definitions and provided a historical context, I’d like to argue that that there are practical and societal benefits from knowing when something we see, read or consume was made with AI and to what extent. It’s also good for human creators too to put a note that their art or even scientific paper is entirely, partially or not-at-all “human made.”

Before we look at the how let’s look at the why and some of the benefits with aknowleding or even citing AI:

Transparency of AI Usage: As consumers for art and idea, I would like to know to what extent a piece of writing or art included AI. Consumers are being unknowingly presented with information and art that was entirely made by AI. This is not necessarily a bad thing, since art, even AI art, has a therapeutic value, and it’s empowering to have various tools help us create. Similar to nutritional stats and ingredients in foods, AI usage disclosures enable us to actively decide to support and consume art that was done exclusively or primarily by AI or by humans. By being transparent about AI being included, we allow consumers of art and information to be more informed about what’s inside what we consume. It also makes is possible to curate what we consume towards the amount of AI in our food, to turn a phrase.

Recognition of AI’s Role/Contribution: Much in the same way that attribution and citation have become common practice in academic and scientific writing, acknowledging AI’s role can help us understand where ideas come from and track references. Part of the challenge with citing AI (for now) is that we do not deem it as an actor or agent like a human writer or publication, so citing an LLM is not entirely obvious. We should cite entities with free will, responability or agent. I also recognize that in the same way we don’t cite that we used Word or Google docs when we deliver a piece of writing or leveraged Photoshop or Figma when we post a photo or graphic, citing the tool seems a bit silly. That said, I think LLM-powered AI’s represent a new class of tools that carry a different kind of contributor role and one that will increasingly prove vital in our human-AI hybrid future. So it makes sense to start acknowledging AI’s role and contribution in some capcity.

Highlighting the Human Creator’s Role: Personally as a writer and artist, I worry that the overease of creating an audiofile or blog post with AI already does and will continue to challenge and diminish the perceived value of my work as a primarily human creator. As counterintutive as this might sound, I believe by citing AI’s role, we are actually highlighting the primary of the human in most creative projects. When we got to a museum and look at painting and sculptures, we see the title, artist name and date affixed. By affixing some label of AI too, we recognize that AI made certain aspects possible but still remain true to the human innovation and spark of imagination involved as well.

As a writer, thinker and creative, I want to be transparent on how I use AI to write and create art and I think it’s a significant part of the kind of world I want to live in too. The rules of how AI should be used and if, when, and how we want to aknowlwedge it remains to be defined. As creators, it’s key to recognize the new role AI can take in our art and find ways to navigate this authentically and purposefully.

Generally if someone wants to read and support human art, disclosures allow us to see this choice point more explicitely. It also gives us transparency and option in order consumption and learning practices.

It’s also time to start disclosing AI’s role in our art and thinking.

Here’s my proposal.

AI Included Disclosure Acknowledgment (AIDA)

The core idea is simple: When you share a piece of art, content or information, you disclose your AI usage. I call this “AIDA” which stands for “AI Included Disclosure Acknowledgment” and is a tribute to Ada Lovelace, a 19th Century female computer pioneer.

AIDA is a simple yet acknowledgement template approach. When you post a piece of writing or share a piece of art, you put an acknowledge or AI citation about your usage of AI. If the art or post was entirely human, you get full credit. If you used AI in small or big ways, you acknowledge it is an easy to understand manner. In practical sense, you cite your AI usage and can use, follow or adapt some basic templates.

The goal of AIDA is two-fold:

• First, to bring more awareness to the changing nature of creativity, authenticity and contribution as it relates to GenerativeAI. See Manifesto.
• Second, to provide simple templates that help creators state what role AI had in their creative output or creative process. See Examples.

There are a lot of hard questions around AI usage). For example, what defines authentic human art when AI had a role in it is creation?

For now, I don’t see this as giving AI credit for what gets created. It’s largley about aknowledging that AI took on some role and contribution in my creation or ideation process and I want to share that when I post, publish or put out artistic work. This gives my audience and my network a clear sense that I am leaning into technology but also want to be transparent and open about it, especially as it relates to LLM-powered LLMs and AI image generation.

I want to make it clear that I still consider human artists as the primary driver and reason for art to exist, yet I think disclosures give me a way to balance using new tools while we figure out how to cite their role as collaborators.

We may not fully know how to navigate this brave new world of AI-hybrid-human creations but I for now I plan on sharing how you got there with an idea or creative project in terms of my general AI usage. In math classes, we were often tasked with more than just finding the answer, which we could look up in the back of the book or use a calculator, but instead asked to show our work.

While I don’t think we need a “show your work” component in most cases, we can start by acknowledging what tools we use and at what steps. It’s a simple as a note at bottom of our posts, in the footnotes, or somewhere in our artist statements or liner notes.

How to get started?

AIDA Templates and Practical Usage

I think AIDA and AI usage attribution is a big idea in need of a manifesto and a lot of activism to grow and share. It’s a topic worth debating more. For now, I’ve created a open source repo with some templates and early ideas here: https://github.com/markwk/aida.

Unlike open source software and creative commons, AIDA is not intended as a license or format for shared ownership or usage, though it could evolve towards that. For now it’s intended to offer transparency as a creator and artist on the role AI tools played in your creative output. Much like your math teacher asking you to show your work, it offers a way to understand how you got there in your creation.

Here a few examples:

AIDA Examples [Written Content Only]

Fully Human Creation

AIDA (AI Disclosure Acknowledgement): Fully Human. The following content was written by me without any assistance from an AI-based system.

Primarily Human Creation with AI Assistance

1. AIDA (AI Disclosure Acknowledgement): Primarily Human. The following written content was written by me with the assistance of an AI-based system (ChatGPT).
2. AIDA (AI Disclosure Acknowledgement): Primarily Human. The following written content was written by me with the assistance of an AI-based system (ChatGPT). Specifically I used it to help me ideate and in revision and editing.
3. AIDA (AI Disclosure Acknowledgement): Primarily Human. The following written content was written by me with the assistance of an AI-based system (ChatGPT). Specifically I used it to help me in some aspects of the pre-writing phase (phrasing, outlining), drafting a few individual points, and various clarification improvements.

Principally AI Creation

AIDA (AI Disclosure Acknowledgement): Principally AI Creation. The following written content was primarily written using AI-based system (ChatGPT). Most of the text and ideas were generated by AI with minimal human editing.

AIDA Examples [For Both Text and Images]

Fully Human Creation

• AIDA (AI Disclosure Acknowledgement): Fully Human. The following written content was written by me without any assistance from an AI-based system. Images were created and edited by me.
• Using Sourced Images - AIDA (AI Disclosure Acknowledgement): Fully Human. The following written content was written by me without any assistance from an AI-based system. Images were sourced from and edited by me.

Primarily Human Creation with AI Assistance

1. AIDA (AI Disclosure Acknowledgement): Primarily Human. The following written content was written by me with the assistance of an AI-based system (ChatGPT). Images were generated using an AI-based system (Midjourney) and edited by me.

Principally AI Creation

AIDA (AI Disclosure Acknowledgement): Principally AI Creation. The following written content was primarily written using AI-based system (ChatGPT). Most of the text and ideas were generated by AI with minimal human editing. Images were generated using an AI-based system (DALL-E) and were unedited by me.

NOTE: These are some just initial starter template ideas. Please fork/comment/share on https://github.com/markwk/aida if you have other examples and better template copy.

Alternatives

For as large of a concern as AI, LLM and Generative represent, I’m quite surprised how little has been written or shared on this topic. I suspect this will have to change. Two alternatives came to my attention as I was completing this post:

Citing ChatGPT as a Reference: While I don’t think this approach really works for me, the American Psychological Association (APA) has created a rubic for citing ChatGPT that looks like this:

When prompted with “Is the left brain right brain divide real or a metaphor?” the ChatGPT-generated text indicated that although the two brain hemispheres are somewhat specialized, “the notation that people can be characterized as ‘left-brained’ or ‘right-brained’ is considered to be an oversimplification and a popular myth” (OpenAI, 2023).

Reference

• OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat

Open-ended Acknowledgement: Rolfe Dlugy-Hegwer shared in Example of how to disclose AI a simpler approach that he appends to his posts and looks like this:

(Disclosure: I ideate and draft content in a text editor, refine it with the aid of artificial intelligence tools like ChatGPT and Grammarly, and revise it to reflect my intended message.)

While I think this is an interesting example and contains the same kernel as what I’m proposing with AIDA, I think it’s too vague and open-ended. I think it provides a good starter for any author using AI tools regularly, but it leaves open an actual acknowledgement or clarity on the time of AI contribution.

Conclusion

Humans are better because of AI, technology, the internet, etc.

Generative AI and other technology tools have changed the nature of work, creativity and even human agency. Among other things, it’s made software engineers, content writers and marketers more productive. It’s changed how we might learn new topics, imagine and get things done.

When you use AI during the ideation, content drafting and final editing, who is the creator? You, AI or some combination?

I’d argue that humans remain the primary creative force behind most art and information, but we are increasingly seeing AI works positioned as human works. This creates major challeges for defining the value of human art and being able to effectively choose to consume human, AI or hybrid creations.

In lieu of some powerful technology that can label AI generated art and information, it is on us as creatives and society to start dislosing AI’s role in our work. We need a new social norm or professional code of ethics that starts with aknowledging the use of AI tool in our creative output, productive work and (to a limited extent) our creative and learning processes. The chief benefits of such AI disclosures are greater transparency (for consumers), clearer recognition of AI’s role/contribution (for creatives), and finally a mechanism that highlights the continuing preeminence of humans as main creator.

This is a challenging topic and one worth debating. Personally, I believe such AI acknowledgements will provide a new social norm and allow us a clearer path forward as creators and consumers of knowledge, creative art and beyond. Generally if someone wants to read and support human art, AI disclosures like AIDA allow us to see this choice point more explicitely. It also gives us transparency and optionality for consumption and learning practices. It brings to the surface what is often hidden and ehlps us navigate a rapidly changing artistic and informational landscape.

It’s tempting (as popular media tends to do) to doomsday that AI is already replacing most workers, including high-paid tech workers. I personally think and believe that for now (as it has in the past) there is an even greater need for technical skills than ever before, specifically specialists with AI skills, models, approaches and tools.

Admittedly, I use ChatGPT and other tools a lot, as I detailed in my 2023 data analysis and write-up on how I use ChatGPT. In that post I pulled my ChatGPT chat log and use some code and tagging to breakdown and reflect on my AI usage. I personally use AI at various stages and steps in my creative processes. I still make and write music myself, though I do use AI to help master and finish music. I use AI Art Generation tools for inspiration and cover art. I use AI tools in various writing tasks from ideation and research to draft phrases, editing and even helping me code and build products. My intention is not to use AI to replace myself as a owner, driver and creative voices in the creative output.

For one of the key areas I use AI for is to help me ask questions. As anyone who works on ambiguous work and creative challenges, a lot of times I get stuck or can’t get started, because I have clarified what I care about or articulated different options or anxieties I’m feeling. Frankly there are lot of lots pre-work challenges that limit my ability to get started and do the hard thing. This is where AI-enhanced reflection can be beneficial and is part of the core value I’m trying to deliver in my app, Stay Reflective, which provides simple ways to get started on a personal, professional or creative challenge and then use AI to generate follow-up questions to think critically or reflect further.

Both fortunately and unfortunately AI has made it easier than ever to make stuff. It’s fortunate because it’s diminished the technical gap between having an idea and actually creating it. This is great and worth celebrating. It’s unfortunate because Generative AI tools have diminished the role of humans as creators and made many of us feel discourage from attempting to create and learn. What is point of working so hard to learn a creative or technical skill or spend time making something “by hand” or by human mind when technology and AI tool can just make something an audiofile, blog post or graphic image so easily?

I believe that AI is a society changer but I still believe there is a role for human creators and thinkers too. Generative AI has made it trickier and trickier to see if a piece of content or art was human or AI made. In world where we want to value and consume human-made art, ideas and information as well as continue to respect the place of humans, it seems time to think about new social norms and societal practices. Like the evolution of citations in science and acknowledge in writing and art, it’s time to start disclosing AI role in our work and creations.

Hopefully AI disclosures like AIDA can help us continue to create and find meaning in creating with or without AI and other technologies!

AIDA (AI Disclosure Acknowledgement): This post was written entirely by me. I leveraged an AI system (ChatGPT-4) for early ideation, editing, minor revisions and improved clarity. I used an AI system (DALL-E & Midjourney) to generate visual elements which I edited and refined.