I'm starting here because I work with practitioners who have principled concerns about AI, who are accustomed to navigating ethical complexity in their field, and who have so far forged a path of good conscience in order to serve the people they care about.
A client recently asked me to review a change she’d made to her website. She wanted my feedback, so I asked: did she want a few thoughts from me via email, or a comprehensive SEO report I could build out using AI in twenty minutes?
She said she wanted my recommendations. She didn’t want AI involved.
Because we’d talked about it beforehand, we both knew where we stood. The pause and the request for consent mattered to her, and it mattered to me that she could make that call.
I’m starting here because I work with practitioners who have principled concerns about AI, who are accustomed to navigating ethical complexity in their field, and who have so far forged a path of good conscience in order to serve the people they care about.
AI is not going anywhere. Exhausting as it may be, it deserves careful thought. I invite you to bring the same intellect, curiosity, and intention you bring to everything else in your practice, because how you think about this can have a real impact on how you operate.
What’s actually happening
The AI industry has real, documented problems. Let’s name them.
The environmental cost is significant.
Generative AI requires dramatically more energy than conventional computing. According to MIT researchers, a generative AI training cluster can consume seven to eight times more energy than a typical computing workload. The numbers at an industry scale are staggering: global electricity consumption by data centers rose to 460 terawatt-hours in 2022, which would have made data centers the 11th largest electricity consumer in the world, between Saudi Arabia and France. By 2026, that figure is expected to approach 1,050 terawatt-hours, which would place data centers fifth globally, between Japan and Russia. (Source: MIT News, January 2025; Organization for Economic Co-operation and Development)
The carbon footprint is growing just as fast. Since 2018, carbon emissions from U.S. data centers have tripled, and data centers now account for roughly 2.18% of national emissions, approaching the footprint of domestic commercial airlines. (Source: MIT Technology Review, December 2024)
And then there’s water. Data centers divert large volumes of freshwater from surrounding communities for cooling, and more than half of the data centers built since 2022 are located in areas where water demand already exceeds supply. In a world where freshwater scarcity is projected to worsen, the infrastructure powering AI is competing directly with the communities around it. (Source: Bloomberg; Southern New Hampshire University, January 2026)
The consent problem is foundational.
Most large AI models were trained on creative work scraped from the internet without the knowledge or permission of the people who made it. The backlash is significant: more than 6,500 artists, designers, and academics signed a public letter demanding the cancellation of Christie’s AI art auction, arguing that AI companies have engaged in mass theft by training on copyrighted material without consent or compensation. At least 16 copyright lawsuits have been filed against nearly every major AI company. In one notable ruling, a district court found that artists may pursue claims against Stability AI, Midjourney, DeviantArt, and Runway for copyright infringement, though the court has not yet addressed whether training on copyrighted works should be protected under fair use. (Source: Brookings Institution, April 2025) The outcome of these cases will shape how this technology develops. Right now, it isn’t trending in creators’ favor.
The human rights implications are eye-opening.
The European Network of National Human Rights Institutions has documented how AI systems threaten the right to privacy through constant data collection and surveillance, exacerbate existing discrimination by targeting already vulnerable groups (facial recognition and language modeling have shown bias against racial and ethnic minorities), challenge freedom of expression through AI-driven content moderation that can suppress legitimate speech, and undermine transparency and accountability when secretive algorithms make decisions that affect people’s lives without any mechanism for challenge or remedy. These aren’t edge cases. They’re structural features of how AI is currently built and deployed.
We are all still learning about challenges and working to respond.
These are some of the most well-documented concerns, but they aren’t the only ones. Some more speculative or harder to quantify include labor displacement, the erosion of creative markets, the concentration of economic power among a small number of technology companies, and the use of AI in military and surveillance contexts. At the same time, the problem landscape isn’t static. Engineers and policymakers are actively experimenting with solutions. China has begun construction on a wind-powered underwater data center off the coast of Shanghai that uses natural seawater cooling and consumes at least 30% less electricity than conventional facilities (Scientific American, July 2025). In the U.S., a company called Span is piloting mini data centers that sit beside homes and draw on the unused portion of household electricity allocations, sidestepping the grid bottleneck entirely (Fast Company, May 2026). These are early experiments, not proven fixes. But they signal that the problems driving concern today are being treated as engineering and policy challenges, not permanent features of the technology. That distinction matters because it enables us to have hope and to stay engaged in pushing these issues in a more ethical direction.
Why this community is right to be suspicious
The practitioners I work with have spent careers navigating a field full of its own extractive patterns: predatory continuing education, scope creep from undertrained practitioners, referral arrangements that put financial interest ahead of client welfare, marketing that preys on people’s anxiety about their health and appearance, and the presence of racism, sexism, and anti-fat bias threaded through the industry itself. Many of you have developed a finely tuned sense for when something isn’t right.
That instinct is what makes you good at your work. And it’s what makes many practitioners look at the AI industry and recognize a familiar shape: a new technology promising transformation while the people closest to it absorb the most risk.
So how do you move forward?
You’re not alone in working through this
I want to offer a practical framework for thinking this through for your practice, but first, it’s worth knowing that this conversation is happening across communities far beyond wellness. Organizers, activists, nonprofits, and creative professionals are all navigating the same questions, and their thinking can inform yours.
Lee Anderson and Oluwakemi Oso, Black organizers and technologists at re:power, wrote a piece in The Forge called “Power, Not Panic” that reframed the question for me. Their argument: AI is a terrain of power, and ceding that terrain to billionaires and autocrats doesn’t protect anyone. It leaves the architecture of the future in their hands. They write that progressive organizers must treat AI the same way they treat policy, narrative, and strategy: as a site of struggle that can be studied, contested, and reshaped. And they make a pointed observation: disengagement doesn’t protect us, it reinforces the systems we’re trying to dismantle.
The Center for Artistic Activism (C4AA) has been thinking through this publicly with their community. After months of internal reading and deliberation, they arrived at a position I find particularly useful: a diversity of strategies can be a strength. Some people and organizations are well-positioned to organize around AI, pushing for regulation, environmental accountability, and data protections. Others need to be using the technology, building with it, applying it, shaping it toward their own ends. C4AA’s position is that both approaches are needed, and that insisting on one correct position while everyone else is complicit or naïve doesn’t serve anyone.
Meanwhile, organizations worldwide are working to develop universally recognized labels for “human-made” products and services. The BBC reported in March 2026 that this certification movement is gaining momentum, particularly in the arts, though AI experts note that agreeing on what truly counts as “human-made” will prove complicated as AI becomes embedded in everyday tools. The instinct behind the movement points toward something economists are starting to articulate. In a recent essay, University of Chicago economist Alex Imas argues that the key question about AI isn’t which jobs it can do, but which services people won’t want it doing. As Ezra Klein summarized in the New York Times, people are “looking at the economy as it exists and asking which tasks A.I. can do; they should be asking which jobs people won’t want A.I. doing.” Imas’s observation: as automation increases, people consistently shift their spending toward work where the human element matters most. They seek out “clothing with a story, food with a provenance, doctors who make house calls, therapists who make them feel seen.” Imas calls this “the relational sector” of the economy, and he expects it to grow, not shrink. Klein noted that economists are broadly skeptical that mass joblessness is on the horizon, and the pattern Imas identifies has played out before: Nespresso machines made espresso effortless at home, and there are more coffee shops and more baristas than ever. The commodity version didn’t replace the human version. It made people want the human version more.
For practitioners in this industry, that’s not an abstraction. It’s a description of the work you already do. If the relational sector is where value concentrates in an AI-augmented economy, holistic health practitioners are already standing on solid ground. The human-made movement isn’t just an ethical stance. It may also be a market signal.
The QuitGPT boycott, a direct response to OpenAI’s deal with the U.S. Department of Defense, has reached over 2.5 million users. Resistance to AI isn’t just individual preference; it’s becoming a coordinated cultural movement with specific economic and political targets. At the same time, Greenpeace advocates for technology built to meet real social and ecological needs, running on renewable energy with full transparency about its environmental footprint, rather than a blanket rejection.
These perspectives don’t necessarily agree with one other. The conversation is live, it’s multi-dimensional, and people with real stakes are arriving at different answers.
A framework for making your own call
With all of this in mind, here’s what I’ve found useful when thinking through whether and how to engage with AI in a practice like yours. This isn’t a checklist that leads to one right answer. It’s a set of dimensions worth considering carefully, so that whatever you decide, you’ve made the decision with your eyes open.
Start with the business picture.
How will your practice be affected by the decision you make about AI today? Can you imagine the impact a year or five years from now? What does “100% human-made” actually mean for your business in terms of the labor needed to sustain it, and how that labor is sourced and compensated, especially against a backdrop where AI is changing the cost structure of nearly every service industry? Taking the time to understand the business impact alongside your ethical values helps you make a more considered decision that impacts everyone you employ and rely upon to operate.
Get clear on your own motivations.
Are you taking a principled stance on AI and accepting the business consequences? Or have you identified a real business need to maintain human connection, meaning the decision to omit AI is primarily about customer trust, brand durability, and long-term practice health? Both are valid, and it could be helpful in building credibility if you speak to the nuance of the decision, should you choose to take a stance.
Consider the practical boundaries.
Innovation is happening fast, which means the decision you make today will need to be revisited again and again. If you allow AI to correct your grammar in email, or you read the AI-generated summary that appears when you do a Google search, you may already be using AI every day. What are the practical tradeoffs you accept in removing AI from your operations? Where do you draw the line, and how do you communicate that line clearly?
Separate your personal choices from your civic engagement.
There’s a meaningful difference between a personal decision not to use AI and participation in an organized boycott. The QuitGPT movement is a coordinated economic action with a specific political target: OpenAI’s relationship with the Trump Administration. That’s civic engagement designed to force accountability. A personal choice to avoid AI may reflect your values, but on its own, it doesn’t necessarily advance the public discourse, support legislation, or exercise the kind of democratic power that shapes how this technology is regulated and governed. To that end, even if you use choose to use AI, you can be involved civically in pushing the industry towards a more sustainable direction.
Recognize that you don’t have to decide everything today.
If you’re feeling unclear or uncertain, that’s a reasonable place to be. You don’t owe anyone a public declaration right now. The vulnerability in admitting that you’re still working it out can bring you closer to your community, where many people are feeling the same way and could use authentic support in moving through it. As C4AA put it: at early stages, when nobody knows which approach will ultimately win, staying open and gathering information is itself a strategy.
Now get specific.
When you’re facing a particular decision about a particular tool, these questions can help:
Where can I actually exercise choice? AI has been embedded in many tools without your input. There are other places where you’re making an active decision to use it or not. Knowing the difference helps you spend your energy where it matters.
What am I using this for, and does the benefit warrant the cost? Not every AI use has the same tradeoff. Transcribing your own session notes with a privacy-compliant tool is different from generating marketing content. The worthwhile question may be whether your specific use makes sense given what you know about the costs.
Who benefits from my use of this tool, and who doesn’t? This is the question this community already knows how to ask, and it is what drives you to offer sliding scale fees or participate in community events. Alongside all other operational decisions, such as where you shop and how you hire, you can apply a community minded approach to the decision of whether and how and when you use AI as well.
What do I disclose, and to whom? This is a live question for any practitioner using AI in client-facing work. How you answer it reflects your values and shapes your clients’ ability to make informed choices. (If you want to dig into this, the disclosure guide elsewhere in this collection, “How to Be Upfront About AI in Your Practice,” walks through it in detail.)
What experiment can I run before deciding? You don’t have to commit to anything sweeping. Try using an AI transcription tool for your own session notes before using anything client-facing. Draft your own newsletter introduction and compare it to an unedited AI version to see what the tool can and can’t do. Use the briefing template in this collection to test a single task with clear boundaries. Small experiments give you real information, which beats both enthusiasm and fear as a basis for decisions.
Where I’ve landed
I use AI in my work. I want to be specific about what that means.
When I started this business, AI helped me develop a voice and tone guide for my brand. It provided the template and structure. I filled in the first draft, fed it examples of my own writing, and we went back and forth until it became a comprehensive document that I now reference every time I write. I wouldn’t have known how to build one before that exercise, and it has been critical in developing quick drafts. When I hired a designer to create my visual identity, I gave them a creative brief that AI helped me develop, and they told me it was one of the clearest they’d ever seen. I like to think it resulted in a better outcome and a smoother process. That kind of strategic scaffolding is what AI does well: it gives me a structure to think inside of, and I bring the vision and the judgment.
On the technical side, AI has changed what I can build and how fast I can build it. I hired a designer for my logo, type, and color palette, and I’m glad I did. That resulted in a visual story I couldn’t have arrived at any other way. But when it was time to build my website, I worked with AI to bring it to life and make it structurally sound. That took a fraction of the time and budget I would have needed a few years ago. When I’m designing a new booking flow or figuring out how to migrate a client’s blog archive off Squarespace and onto a static site, AI gives me the collaborative problem-solving of a senior engineer and speeds up execution dramatically.
For client work, AI is what makes my pricing possible. I’m building a comprehensive marketing analysis service where AI handles the bulk of the production, referencing best-in-class tools and research, while I steer the strategy, ask questions along the way, and make sure the final product reflects what I actually know about the client. Without AI, that service either doesn’t exist or it costs three times as much.
The same principle can apply to you. Whether it’s developing an updated business plan, pressure-testing a pricing change, building a brand voice you can actually write from, or drafting content that sounds like you and not like a template, these are tasks where AI can do the heavy lifting while you stay in creative control.
That’s the honest math. Without these tools, I couldn’t deliver the scope and quality of work I deliver at the price points I charge.
I pay attention to which tools I use and under what terms. I choose providers with stronger commitments to privacy, transparency, and data practices wherever I can. I don’t use AI for anything that requires knowledge of a client’s specific situation that should stay between them and me. When it comes to work I do for clients directly, I offer disclosure and ask for consent before AI touches anything I’m delivering. The conversation at the beginning of this piece happened because I’d made that commitment.
I’m building a custom booking experience for a client now. I consult with AI about how best to build it, but there’s no reason to embed AI into the product itself. That means this provider will have a highly sophisticated system that is completely AI-free. The aim is to cut her admin time in half without moving her onto the industry’s standard booking platform. Her system will just work, without subjecting her to whatever subscription fees, price changes and redesigns the bigger companies push through to keep up with the AI race.
I don’t have a clean answer to the ethical tension. Choosing tools carefully doesn’t resolve the environmental costs, the consent problem, or the labor questions I laid out earlier in this piece. I use these tools with that awareness, and it shapes what I choose, how I use it, and where I put my energy outside of my work. It’s not enough, but it’s not nothing.
I know some people will read this and decide not to work with me. That’s okay. The disclosure gave you what you needed to make that call, and that’s as it should be.
For everyone else: I don’t think there’s one right answer to the question of AI in your practice. I think there are thoughtful ones and reflexive ones. AI is generating a lot of anxiety at the moment, but you can treat this the same way you treat everything else: with curiosity, with care, and with a willingness to revisit as you learn more.
The decision gets easier when you’re not making it alone. If it would help to think through this together, I’m here.
Sources Referenced
- Explained: The Environmental Impact of Generative
AI — MIT News, January 2025
- AI’s Emissions Are About to Skyrocket Even Further — MIT Technology Review, December 2024
- AI’s Environmental Impact — Southern New Hampshire University, January 2026
- AI and the Visual Arts: The Case for Copyright Protection — Brookings Institution, April 2025
- Key Human Rights Challenges — European Network of National Human Rights Institutions (ENNHRI)
- Human-Made Certification Movement — BBC News, March 2026
- AI, Energy, Environment and Democracy — Greenpeace International, April 2026
- Power, Not Panic: Why Organizers Must Engage with AI — Lee Anderson and Oluwakemi Oso, The Forge / re:power
- Six Thoughts About AI and a Workshop — Center for Artistic Activism (C4AA)
- Why the AI Job Apocalypse (Probably) Won’t Happen — Ezra Klein, The New York Times (see also: What Will Be Scarce? by Alex Imas)
- You can put a data center at your house — but who really pays? — Mark Sullivan, Fast Company
- China is putting data centers in the ocean to keep them cool — You Xiaoying, Scientific American