Manifesto of an AI refusenik

What a world we live in! You don’t need to read long articles or emails; AI will summarize them for you. You don’t have to sift through numerous sources; AI will tell you what is known about a subject. You don’t have to cultivate your client relationships; AI will write letters to your customers for you. In fact, AI will write… your online content, your school papers, your performance reviews, your annual report, your video scripts, your next novel… for you. You don’t have to hire or develop media skills; AI will create a custom graphic or photograph or video for you. And… the tools and capacities of “generative AI” * keep getting better and more widely available. Imagine what may be possible soon.

By taking over all this mental labor, AI frees up your time. By outsourcing it all to a computing model, AI lowers your costs. Who wouldn’t want more time, and more money to spend?

But the hype ignores a long list of significant harms – harms to our accurate understanding of each other, harms to people’s intellectual growth, harms to our economy, harms to our planet.

As a writer, I will not use GenAI tools for any purpose other than grammar and spell check (run by Microsoft Editor, not Grammerly). As a professor, I do not use AI tools to build my courses, and I actively discourage my students from using AI to support their learning. As an entrepreneur, I have enlisted a professional to create graphics, and to support my web site design. It doesn’t matter to me how “good” AI tools claim to be; the offenses are simply too great.

Consider these harms…

GenAI produces inaccurate summaries and wrong answers. A GenAI product does NOT: think, analyze, synthesize, problem-solve, counsel, or befriend, to name just a few of the human cognitive and interpersonal attributes that people use to talk about GenAI capabilities. Tools using GenAI to process questions and produce answers (in text, video, or audio) are egregiously – and even dangerously – wrong often enough to be worrisome; and many users don’t have the capability to distinguish errors. GenAI is at its core only a prediction engine, albeit a very powerful one. It’s show without substance, and can’t be fully trusted. Numerous studies and tragedies demonstrate this.

GenAI compromises learning processes by discarding nuance and allowing critical thinking skills to atrophy. Learners who only see summarized material miss out on being exposed to detailed arguments and deep thinking. Learners who let AI produce early-stage drafts don’t learn the core skills of their field. Learners who don’t have to do their own synthesizing and don’t have to stare at blank pages until they can articulate what they think are losing real insights that can come only from the workings of a human mind with unique experiences and knowledge. 

GenAI is unethically built on stolen and copyrighted resources and undermines the value of human creativity. GenAI companies now make money for their shareholders without compensating or accurately referencing the creators whose work has been scraped to train these systems. Indeed, they never asked for permission to use these copyrighted materials and intellectual property.

AI is creating serious risks in our economy. Many economy-watchers warn about the way that AI companies are propping each other up with circular investments and purchasing contracts, and about the deep financial commitment so many organizations are making in AI tools. Should AI not live up to the hype, the market will be in deep trouble. Too many organizations extol short-term bottom-line impacts without considering the consequences of outsourcing a substantial amount of human work to computers. What will happen when all those jobs are lost? What will happen when AI agents are not up to the tasks that are envisioned? The futurists who are examining that potential downward spiral paint a terribly bleak picture.

AI is compounding ecological and social justice issues by building energy-devouring and water-thirsty data centers, often in marginalized communities. People are waking up to these issues and are beginning to object to data centers being built in their cities and towns. Where data centers are built: electricity costs rise alarmingly; off-the-grid power generation produces substantial air and noise pollution; and the draw-down of precious water resources increases to unsustainable levels. We need a more proactive analysis of these issues before spinning up huge data centers.

AI tools gather personal information and have few protections to safeguard individual privacy and security. We are allowing companies that have no guardrails to collect, use, and sell OUR data without our knowledge or permission. Worse, these tools are invading schools, putting our children’s privacy at risk. Too many people are unaware of the end-stage risks and unable to opt out of engaging with these systems.

In the hands of nefarious actors, GenAI creates gigabites of slop that pollute the internet, including materials that are meant to spread lies and rile emotions; worse, GenAI is being used to create and circulate porn. Who knows how much of this slop is being fed back into the machine, further degrading AI outputs.

I am lucky to be able to refuse to use GenAI to create content for me when so many people are required to make use of it, or are feeling intense pressure to ride what is advertised as the wave of the future. AI is preemptively embedded in many software tools, and I sometimes have to dig deep into settings to turn it off (if that option even exists). I would never want to give up the joy I get from research, hard thinking, and crafting words into paragraphs.

With a lifetime of experience in teaching and training and deep knowledge of learning dynamics, I am very concerned about the adoption of AI tools and “AI literacy” in schools and universities; I believe it does more harm than good. I am beyond dismayed by suggestions that AI tools can replace teachers, and can replace the training department in organizations.

I continue reading smart pundits on both sides of the debate around the use of AI, especially in educational technology, and I find the arguments made by cautionary analysts far more compelling. We’ll see if these warnings get tempered as time moves on, but without broad action to address harms, I suspect they may get more forceful. We should demand in-depth open debate about AI and its uses, and everyone should be able to easily opt out as their consciences dictate. We should be actively considering regulations in order to prevent powerful people from making decisions that are not in our best interest.  

We are meant to build a world in which humans can thrive. While some AI advocates tout wondrous potential contributions, they gloss over the costs of getting us there. There’s an old adage that if something sounds too good to be true, it probably isn’t good or isn’t true. Many of the companies pushing AI are the same ones who gave us social media, which they managed to turn to disaster. Let that be a lesson.

* Generative AI refers to artificial intelligence tools that create text, images, videos, audio, and software code by using statistical analysis of source material to generate coherent responses to prompts (e.g. ChatGPT, Copilot, Gemini, Claude, Grok, Midjourney, DALL-E). Other categories of AI have been described as predictive, assistive, agentic, and conversational.

For a taste of the cautionary arguments:

Demystifying Anthropic and ClaudeAI with Gideon Lewis-Kraus and Chris Hayes, Why is this happening? podcast, March 2026

Is AI Making Us Lazy? Cal Newport, June 2025

LLMs: Dishonest, unpredictable and potentially dangerous, Gary Marcus, June 2025

AI is Homogenizing Our Thoughts, Kyle Chayka, The New Yorker, June 2025

More than 200 environmental groups demand halt to new US datacenters, Oliver Milman, The Guardian, December 2025

The 2028 Global Intelligence Crisis: A Thought Exercise in Financial History, from the Future, Citrini Research, February 2026

Privacy in an AI Era: How Do We Protect Our Personal Information? (2024) Stanford University Human-Centered Artificial Intelligence (full white paper also available for download)

AI Slop: Last Week Tonight (2025) with John Oliver (a comedic presentation of a serious analysis)

For critiques of AI, follow: Audrey Waters (ed tech), Gary Markus (general AI commentary), John Warner (writing and teaching writing, see also his book More Than Words); and tons of links compiled by Cate Denial: Against Generative AI