This post is part of a series on AI and my journey using it as a multimedia storyteller and journalist.

There are so many ways to use generative artificial intelligence (AI). Some are relatively innocuous, and some are pretty controversial. We’ve all heard stories of people getting caught using AI in questionable ways: from authors using it to add full passages or quotes to their books, to people “vibe coding” websites with security risks, to students using AI to write papers or do their own homework.

In the most unfortunate scenarios, the use of AI has severely damaged people’s credibility. The rapid adoption of generative AI tools in the past few years has also led to finger-pointing and confusion when it comes to who is and isn’t using AI in their professional careers and to what degree. It also has people questioning the ethics behind using AI at all.

This is why it’s important for journalists and media professionals to be transparent about whether they use AI and how. Even though many media organizations are starting to create their own AI policies and guidelines, that may not be enough for freelancers, job seekers and reporters at organizations with vague or nonexistent AI rules.

That’s why I’ve created a personal AI policy and am publishing it publicly. It has two major components: a short AI disclosure statement to publish on my social channels and blog, and a more detailed set of guidelines broken down by use-case category.

I highly recommend every creative, writer, and media professional do this. Then be honest about it and stick to it in order to minimize confusion about your work going forward.

To come up with the policy, I wrote down exactly how I might use AI professionally by use-case category and wrote my hard no’s for it. Then I summarized all of that in the disclosure statement.

TL;DR: Here is the full disclosure statement:

How I use AI

I’m passionate about the ethical and transparent use of AI. I use AI as a tool and reference. I do not use AI to replace my own creative or professional authorship unless disclosed.

Specifically, unless appropriate and transparent, I do not use AI to author original writing (articles, blog posts, stories, screenplays, etc.), to use as an authoritative source, or to misrepresent my skills. I do not use AI in violation of client or employer policy, including inputting deliverable copy into AI tools. I do not use AI to author employment application materials (i.e., cover letters).

If appropriate, I may use AI for assistance with functional tasks (i.e., formatting, spelling and AP style), creating functional content (i.e., email drafts, marketing content, social media assets and administrative documents), brainstorming, or troubleshooting/debugging. I also use AI as a creative medium and tool to help execute my own audio-visual ideas, usually in a highly transformative and expressive way (i.e., short films). I manually review, edit and test AI output.

The statement’s format is:

  • Paragraph 1: Quick summary of personal AI philosophy, acceptable use and personal limits
  • Paragraph 2: AI hard no’s and personal boundaries
  • Paragraph 3: AI uses

And here is the breakdown by use case:

(A note: Generative AI is still evolving, and things can change. I’m happy to reconsider some of these. Entries with an asterisk (*) are ones that I could see potentially being amended in the future. If I do amend them, I will update this page and policy accordingly. Also, the items on this list are guidelines for if or when I use AI. I don’t use AI in these ways all the time. In the majority of cases, I may not use AI at all!)

Overview

I use AI as a tool and a reference. I do not use AI to replace my own creative or professional authorship.

How I Don’t Use AI

  • Authorship, voice or prose
  • Authoritative sourcing
  • Inputting confidential information

How I May Use AI

  • Resource and reference (search, exploratory research, and sourcing ideas)
  • Functional tasks (numbers, math, spelling and grammar)
  • Functional content
  • Formatting
  • Planning
  • Brainstorming and ideation
  • Troubleshooting, rubber ducking and spot-checking
  • Editorial, illustrative and supplemental graphics
  • Efficiency apps and tools for SEO and discoverability

Best Practices

  • Always double-check the company or client AI policy and adapt personal policy accordingly.
  • Always carefully review AI output (fact-check, edit, test, bulletproof, scrutinize, etc.).
  • Always be transparent and don’t deceive.

Writing + Editing

AI is a good tool for cleaning and editing copy (functional tasks) in the absence of a human editor, but not for authorship or prose. When working with clients or an employer, I rarely, if ever, use AI due to having an editor to work with, more time to produce the work, and varying or nonexistent company/client AI policies to abide by.

How I Don’t Use AI

  • Authorship and professional work: This includes using AI to author or compose professional articles, blog posts, scripts, screenplays, manuscripts or any professional writing perceived as being original work I authored myself, unless otherwise disclosed and done transparently. This also includes drafts. I don’t use AI to write “drafts” of my work.
  • Prose: This includes using AI to enhance or rework my writing or writing style outside of spelling, grammar and AP style.
  • Sharing confidential copy: Knowing that some AI tools monitor, collect or store input, I do not input any content or copy that is to be later submitted to a client or employer.

How I May Use AI

  • Personal copyediting: When I write for myself (i.e., blog posts), I lack an actual editor. I may use AI to help only with AP style, grammar, spelling and fact-checking. Here is the prompt I use:
    • Edit the following for AP style, grammar, and spelling. Separately, point out any factual flags and any issues in a list. Do not rewrite the original, except for AP style, grammar and spelling.
    • I then review the results, judiciously going through any list of suggestions. I don’t allow AI to write or draft any copy other than to correct style, grammar and spelling.
  • Administrative and functional content: This includes social media, marketing, emails, and letters.
  • Reference tool and research: Using AI like a reference tool, much like a dictionary, atlas or encyclopedia, or to confirm information, methods or procedures.
  • Self-critique and bulletproofing: Asking AI to find holes in an argument or methodology as a stress test.

Research + Background

AI can be a good jumping-off point for research, especially as an alternative to using decaying search engines, but it is a horrible principal source or definitive source for, well, anything.

How I Don’t Use AI

  • Authoritative sourcing: I may use AI as a jumping-off point or lead. I don’t use it as a main, primary, authoritative, or final source.
  • Copy/paste: I always fact-check, scrutinize and review AI claims before publishing them. I don’t copy and paste directly from AI without proper diligence and scrutiny. (BTW: If copying and pasting information into notes, which should be avoided, it’s a good idea to mark the note as being from AI so that you don’t accidentally mistake it for your own wording later on.)

How I May Use AI

  • Google alternative: Google feels less reliable these days, even for the most basic search queries. AI has been a useful way to find links to answers to general questions. It sometimes does a better job pointing in the right direction. Like all AI output, AI should only be a push in the right direction, not the final word. This is useful for finding sources, resources, links and studies.
  • Tough questions and synthesis: Asking tough questions whose answers need to be broken down or talked through. Using AI to get a synthesized layperson’s breakdown of complex ideas as a starting point for deeper research can be useful.
  • Education and training: Getting a general understanding of a concept and creating quizzes and challenges to practice. This has been particularly useful in learning foreign languages.

Data Analysis + Math

AI can help with some data analysis tasks. When creating data-driven products and deliverables, however, I prefer to manually scrutinize and bulletproof data.

How I Don’t Use AI

  • Data collection: I don’t trust AI to simply give me accurate data or datasets. I prefer to find datasets from other traditional sources. I may use AI to direct me to these sources, but I don’t simply trust or use AI-generated datasets without verifying the data.
  • Copy/paste: I do not use or publish AI data, calculations or results without fact-checking and bulletproofing.
  • *Data visualization: I don’t use AI to create data visualizations. This is mostly because I don’t see how I would, or the benefit of doing so. I prefer to use traditional tools like JavaScript. Fact-checking AI-generated data visualizations also seems like it would be a pain.

How I May Use AI

  • Cleaning: Quick formatting and data cleanup, for example:
    • Convert this list of 500 cities into AP style.
    • Format this data into a table with the following columns.
  • Bulletproofing help: Help double-checking math, methodology, spot-checking, etc. This would be in addition to using non-AI techniques.

Coding + Programming

While AI can streamline the programming process, I do not use AI to simply write code for me that I don’t understand, a part of what some call “vibe coding.”

How I Don’t Use AI

  • *Vibe-coded projects: This refers to the idea of trusting AI to code or program an entire product or project simply from a prompt. The caveat here is that I’m not completely against this, and if I did do this, I’d want to be transparent about it, avoiding claiming authorship. As of now, I’d just rather not: I have learned to code in several programming languages and enjoy the learning process. Using AI to create an entire product and then copy and paste code without knowing how it works or how to review it does not appeal to me and seems risky. While it may be useful to use AI to generate bits and parts of a large project, I do not use it to create complete products, as of now.

How I May Use AI

  • Planning: Using AI to blueprint and plan a coding project.
  • Troubleshooting: “Rubber ducking” and debugging my code, or asking “how-to” questions that I would otherwise ask a Stack Overflow or Reddit community.
  • Conventional/structural code: Using AI to create reusable bits of code, boilerplate code, or frameworks for efficiency.
  • Education: Learning new or more-efficient programming concepts, methods, approaches, tools, etc.

Arts, Graphics + Visuals

I tend to distinguish standalone art from supplemental graphics and visuals. When using AI to generate any kind of art, I typically act as an art director, not a client. AI-generated art tends to require a great amount of human-guided creative direction and a vision. Transformativeness becomes the key concept here.

How I Don’t Use AI

  • Professional writing: I don’t use AI to compose professional or creative writing I plan to share or publish under my authorship (blogs, stories, scripts, manuscripts, articles, etc.), unless the AI use has been clearly disclosed.
  • Informational graphics: I generally don’t use AI to publish standalone infographics or informational content that is not illustrative or supplemental.
  • Art: I don’t publish raw AI generations as my own work of art unless it has truly been transformed, altered, edited, composed, etc., into a larger work that fulfills a distinct human-generated vision (a story, style, vibe, aesthetic, etc.). The exception here is for personal or private use, like creating a screensaver.

How I May Use AI

  • AI production and AI-assisted storytelling: I use generative AI as a tool to create larger works rather than just the generation itself. Examples of this include making AI movies. Instead of prompting for a completed film, the process involves acting as a creative director to bring dozens or hundreds of generated assets (images, clips, music, etc.) together to fulfill one human-conceived vision.
  • Assets and illustration: Creating graphics, clips or audio for illustrative or supplemental purposes, such as a lead image for a blog post, or background music. I tend to use this for marketing and social media materials, usually as an alternative to using stock images.

Brainstorming + Ideation

Personally, I prefer to brainstorm things myself. It is one of my strengths. If I do use AI in brainstorming and ideation, it is generally as an assistance and not a final source.

How I Don’t Use AI

  • Copy/paste: I don’t simply copy and paste ideas without editing, scrutinizing and fact-checking.

How I May Use AI

  • Story ideas: Particularly for blog posts or social media content. I tend to avoid with professional writing (articles) and do not use it at all for creative writing ideas.
  • Getting started: Getting started with a story, including source ideas, background information and leads.
  • Optimization: Particularly when dealing with marketing or social media copy, AI helps with optimizing assets, meta tags, headlines and marketing content for the best SEO and attention-grabbing results. I don’t often rely on this, as I have years of training and experience with SEO.

Job Applications + Career Development

Given the tedium, repetitiveness and competitiveness of job searching, it’s tempting to use AI to streamline the process. However, as a writer, I draw a pretty hard line here. Just to avoid any confusion, questions or problems going forward, I avoid using AI in my job application materials. Instead, I may use AI tools to optimize my professional channels to best convey my skills online.

How I Don’t Use AI

  • Professional/application materials: I do not use AI to generate any job application material, including resumes and cover letters.

How I May Use AI

  • Optimization tools: I may use AI tools to optimize my professional social media content with the best keywords, headlines, etc., for SEO, visibility and clarity. LinkedIn, for instance, offers premium tools that may suggest profile descriptions or tags/skills to add to my profile.

I don’t use AI all the time or for every professional project I have. These lists include possible usage scenarios, not necessarily things I do on a regular basis. All of these are flexible and can change if it makes sense. Ultimately, my goal is to make sure I am using AI ethically, transparently and without being misleading.