Let’s retire fake outrage and feigned confusion

Lately, we’ve been seeing quite a lot of disingenuous arguments regarding AI. And while confusion and anger get engagement, these issues are too serious to descend into intentional misunderstanding and false offense. However, the conflicts aren’t all manufactured, and they aren’t just happening online. People we know are either sincerely misled or else sincerely confused about the basics of the AI discourse. We want to have a more productive conversation about AI moving forward.

But we can’t have an honest conversation unless the participants agree on what it is they’re talking about. So, in this blog post, we’re going to go over…

• A few definitions (including “AI”)
• What data centers are, how they operate, and why they’re being built
• What the main concerns over data centers and generative AI are
• How we can talk about problems with AI with the aim of reaching understanding

We’re also going to touch on…

• Disingenuous questions
• False comparisons

So if you’ve been wondering about these issues or trying to find out how to have more productive conversations about AI, read on.

What is AI?

Though AI stands for “artificial intelligence,” (it’s sometimes also called “augmented intelligence”) it isn’t technically an intelligence. In the same way that “autopilot” in cars isn’t an actual autopilot, large language models aren’t actually AI.¹ At least, not yet. LLMs analyze linguistic data and make predictions based on those analyses. They ‘recognize’ patterns in speech and writing so well that they can generate their own human-sounding text. Proto-chatbots existed as far back as the 1960s (such as ELIZA), but bots as we know them today didn’t appear until the late aughts. Cleverbot arrived in 2008 and began making a splash online.²

Learning models were developing quickly all over the world, from England to Japan. In 2024, a news story about a computer program from a Japanese bakery started gaining traction. ZME Science reported³ that a decade prior, BakeryScan—a neural network model⁴ developed to identify pastries—started making waves. The article says that a doctor in Kyoto saw the potential to use this model to identify cancer cells. A few years later a prototype appeared in hospitals.

Eventually, this lead to generative AI models like OpenAI (ChatGPT), that could generate content that was virtually indistinguishable from a disinterested human writing to a template.

Yes, artificial neural networks could make life easier

When most people praise AI, they’re thinking about its potential. These models could eliminate most of the “busy work” that people—especially people who can’t pay for convenience—have to do. Here’s one example. If you need a mechanic, a well-trained and accurate model could compile information about local auto shops, including:

• reviews
• years in business
• complaints
• average rates
• and other important factors

All of this, in seconds, allowing you to compare them side-by-side. That would save quite a lot of time and frustration. Or an AI tool could make grocery shopping simple. It could:

• scan all the sales at local grocery stores
• compare that to your weekly grocery needs
• consider the time you have to shop
• factor in price changes from week to week

Then, it could generate an almost-instantaneous shopping itinerary with a route that could save you time and money.

But, at least so far, legitimately useful AI tools like these aren’t freely available to the average- or low-income person. Instead, AI is being leveraged against everyday people with applications like dynamic pricing (which can track what you buy and increase the price on items you need most). So while we acknowledge that these tools have the potential to change our existence for the better, what the average person is actually facing are more problems from AI than solutions.

What are the supposed problems with AI?

What started out as a technology that offered fun interactions with chatbots and life-saving medical applications quickly cascaded into a cause for concern. We’ll discuss data centers in a moment, but for now let’s take an overview of the problems with AI itself. The primary concerns making headlines are the loss of jobs and the loss of skills. Let’s break those down.

Job Loss

In 2025 alone, AI implementation at workplaces caused 54,694 layoffs, according to this report.⁵ The Word Economic Forum predicts that by 2030, 92 million jobs will have been eliminated.⁶ This would be great news if the United States and other affected countries had guaranteed universal basic income, as it would allow many people to do meaningful, human-powered work regardless of whether or not it paid well. But in our current economic reality, those numbers are devastating.

Skill Loss

Professional skill loss due to AI

But what if the economic devastation wasn’t so severe? Would AI still present problems if no one lost their jobs and homes? Well, there are still real concerns with the human cost of relying on AI. “Skill atrophy” due to AI is a particular worry when it comes to doctors’ abilities. In 2025, The Lancet published a study that suggested a reliance on AI-assisted diagnostics could result in a loss of ability to correctly diagnose patients without AI. That is to say, diagnosticians could see their skills decline as they use AI assistance.⁷ We’ll need more rigorous and well-controlled studies to know if this is a threat, but in the meantime it is a cause for concern.

Developmental skill loss due to AI

But it isn’t just diagnosticians we need to keep an eye on. Psychology Today writer Timothy Cook argues that, though offloading cognitive tasks to AI can lead to skill atrophy in adults, it will prevent those skills from ever being formed in children. He describes this as part of the AI audit problem. That is, adults who have learned how to do these tasks and are simply looking to save time still have the capacity to examine AI’s results and see if they’re correct.⁸ But children, who have not learned how to do the task on their own, will not have the knowledge or experience needed to perform this audit.

AI dependence may cause adults to lose skills.
Children will never learn them.

And with more and more students relying on AI, we could be seeing the beginning of a loss of skill far more dramatic than the one we’ve been part of for the last couple of hundred years. This isn’t fear-mongering or pearl-clutching on the part of a handful of sulky, unemployed people. It’s not the rant of an old man shouting at clouds. This is a real potential danger that we should be diligently monitoring and investigating.

What’s the deal with data centers?

Even if we somehow dealt with the loss of jobs, and it turned out that the fears about skill atrophy were overblown, there’s one more major problem with AI proliferation. If we want to have a sincere and productive conversation about AI, we need to consider data centers.

Data centers are not new. But their numbers have risen sharply with the demand for bitcoin and, especially, generative AI products. Structurally, they’re similar to supercomputers; buildings filled with thousands of servers performing calculations. However, while a supercomputer usually focuses its resources on a single computation, data centers perform thousands of separate computations simultaneously.⁹ This article by Forbes outlines the different types of data centers.¹⁰

The problems with data centers

This expansion comes at a cost. While proponents are eager to focus on the temporary construction jobs and handful of long-term jobs at the data centers themselves, residents are fighting to keep these projects out of their communities. Surges in energy costs, the co-opting of millions of gallons daily from local water sources, and the increase in pollution that come along with data center construction are, unsurprisingly, unappealing. Now that word has gotten out about what some residents are suffering, worry and resistance are growing. This public push-back has led developers to add some sustainable features to future projects, but with the long history of large corporations sickening and killing U.S. citizens, it’s easy to see why many people are still deeply skeptical.¹¹

Further, with the increase in energy usage due to AI applications, all that new energy has to come from somewhere. The construction of multiple nuclear power plants (as well as an unknown number of gas and coal plants) is being planned across the country. And while Meta uses phrases like “energy leadership” and “reliable and firm power,” people become defensive when these innovations arrive in their backyards.¹²

Now we’re on the same page about what AI is and what the major causes for concern are. But before we can have a productive conversation about AI, we need to address a few of the most common disingenuous questions, concerns, and talking points.

Disingenuous questions and arguments

You can’t have a productive conversation about AI with these

Often, when someone on the internet says they’re against using AI, a percentage of comments will say something like, “But AI has medical applications!” Or, “Do you really want people to die of cancer?” These responses may be well intentioned, or they might be badly used rhetorical tools to win an argument.

What do we mean by disingenuous? Here’s an example that’s been around for decades:

Bette: “I don’t want to eat a bunch of chemicals on my fruit.”
Bob: “Well actually, everything is made out of chemicals.”

If Bob responds to Bette, knowing full well that Bette meant artificially created and added chemicals, then Bob making a disingenuous argument. Of course, if Bob didn’t know that’s what Bette was referring to, that’s okay. That’s why it’s important for all of us to ask follow-up questions instead of assuming we know what’s being discussed and jumping in with an objection.

False assumptions about AI

Here’s an example we’ve heard regarding AI:

“Isn’t the cost of the environmental destruction AI is causing worth the medical advancements?”

The problem with that question is that few people are unhappy with the existence of data centers that provide computational power to medical applications. Genuinely sustainable solutions can be found and implemented for AI usage designed to save lives. But most people are unwilling to risk poisoning their local water supply and soil just so someone else can earn money through bitcoin or so multi-billion dollar corporations can offer cheap generative AI and cloud computing. The question is comparing apples and oranges.

Racism, classism, and ableism around AI

Some of these arguments go beyond being disingenuous or ignorance-based, however. Recently, we’ve seen some pro-AI arguments that can only be described as racist, ableist, and classist—whether or not that was the intention.

Here’s one that we’ve seen several times:

“But AI is the only way disabled/Black/disadvantaged people can compete!”

While it’s possible this sort of thing is said with good intentions, the racism and classism inherent in it is impossible to ignore. And yet it isn’t new. In 2024, NaNoWriMo faced backlash after stating that condemning the use of generative AI had “classist and ableist undertones.”¹³ Many long-time supporters of the organization seemed to find that statement itself classist and ableist, saying that it implied disabled and poor participants needed help from AI software to write. While there is a larger and more nuanced discussion around the fact that people facing socioeconomic disadvantages are, obviously, disadvantaged when it comes to paying for feedback and finding free time, many found NaNoWriMo’s broad statement cloaked in politically correct language offensive and cowardly.¹⁴

It isn’t a lack of creativity or ability that disadvantages disabled and lower-income creators. It’s a lack of access to editing and publication services. A lack of time due to working excessive hours or losing ‘working’ time to illness. And AI could help break down those barriers. But conflating the use of generative AI to create the end product and using AI tools to gain access or reduce barriers is not helpful.

Possible advantages for the disadvantaged?

How could AI level the playing field for lower-income people? Addressing that is, unfortunately, beyond the scope of this post. However, we’re working on another blog that addresses the topic of AI in art and writing in greater detail. In that one, we’ll look at generative AI in the context of the history of technological advancement in the arts and as a means of eliminating gatekeeping in the literary and music publishing industries. If you’re interested, subscribe to our mailing list or follow us on social media to be notified when that post is live.

• Instagram
• LinkedIn
• Facebook

In the meantime, how can we have a productive conversation about AI? We’re glad you asked.

How we can communicate effectively about these issues?

There are a few very simple ways to make the conversation around AI more productive. First, let’s stop using “AI” in such a broad, all-inclusive way. If you mean generative AI, say that. If you mean AI-powered translations, say that. And if someone says “AI” without any further context, ask them what type of AI application they’re talking about and accept their answer.

It’s also important to remember that many, many things have been retroactively labeled AI. Voice changer apps that have existed for the last 20 years are now being referred to as “AI,” even when no AI has been implemented. Filters that were available with the original Adobe Photoshop released in 2000 are now being called “AI.” The rush for companies to jump on the AI bandwagon has made the language around these topics unclear, so we need to pause long enough to ask a few clarifying questions and make a few qualifying statements.

Keys to having a productive conversation about AI

When you’re talking to someone, either online or in person, assume ignorance, not malice. If you’re talking to someone for the first time, and they appear to be making disingenuous arguments, respond from the point of view that they’re simply uninformed. Remember that the language around AI has been intentionally muddied, and plenty of so-called AI evangelists are spreading misinformation.

Remember too that most people don’t want to destroy their local or global environment, lose their livelihoods, or increase the bloated wealth of the 1% at their own expense. If you seem to disagree with someone over the problems with AI or data centers, the most likely reasons are miscommunication or misinformation, not a misalignment of beliefs.

We believe that the goal of conversation should be mutual understanding, not victory over an opponent. If all people leave the conversation comprehending the points of view of their fellow interlocutors, that’s a productive conversation, whether it’s about AI or any other topic.

---

Have you had genuine discussions about AI, either with people you know or with strangers on the internet? Does this guide seem helpful? And is there anything that we left out? Let us know in the comments below.

---

As LeVar Burton always said, “Don’t take my word for it!”

1. Why LLM is Not AI Yet ↩︎
2. This article, written in 2010, feels a little creepy thanks to lines like, “Chatterbots have been known to keep people interested, and keep them coming back to talk again and again.” Cleverbot Chat Engine Is Learning From The Internet To Talk Like A Human ↩︎
3. The Unlikely Story of How a Pastry AI Came to Be Used to Detect Cancer ↩︎
4. In computing, a neural network is a model that takes some inspiration from actual neural networks found in organic lifeforms. No actual neurons are involved. ↩︎
5. 71,321 Job Cuts on Restructurings, Closings, Economy ↩︎
6. The Future of Jobs Report 2025 ↩︎
7. Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy: a multicentre, observational study ↩︎
8. Adults Lose Skills to AI. Children Never Build Them. ↩︎
9. Supercomputers versus Data Centers ↩︎
10. What Is a Data Center? ↩︎
11. Data Centers, Pollution, and the Communities Left Behind ↩︎
12. Meta Announces Nuclear Energy Projects, Unlocking Up to 6.6 GW to Power American Leadership in AI Innovation ↩︎
13. NaNoWriMo Organizers Said It Was Classist and Ableist to Condemn AI. All Hell Broke Loose ↩︎
14. NaNoWriMo’s AI Stance is a Blow to Disabled Creators ↩︎

Feature image by kiquebg