AI: Love It, Hate It — It Is Probably Here to Stay

First, a few things. I am not an AI expert. I am just a person. I am probably more computer savvy than most, but I am far from the top when it comes to holistic knowledge. I know what I know, and I do not know what I do not know. I can only speak from my lot in life. That lot is determined by what I see, read, observe, and think. I am not married to any one way of thinking. I am open to new ideas and different points of view. This is where I stand, and what I think about this topic right now as I am typing what will probably be a very long very boring rant. God save me from those I am about to piss off, because from what I have seen, some corners of the anti-AI crowd seem to enjoy a good character assassination almost as much as they enjoy the argument itself.

At its core, AI is a tool. Like virtually every major tool humanity has ever invented, it can be used in positive ways, negative ways, stupid ways, destructive ways, and incredibly useful ways. The technology itself does not magically determine which one of those happens the people do. That does not mean we should ignore the consequences of a technology simply because it is useful. I feel we should do quite the opposite actually. We should absolutely talk about copyright, privacy, energy consumption, misinformation, job displacement, consent, data collection, and how these systems are trained and deployed. Those are legitimate conversations. What I have a harder time understanding is the idea that the technology itself must therefore be rejected completely.

We have been through versions of this before. The printing press, electricity, typewriters, cars, the internet. The list goes on and on but for the sake of brevity we will stick with the internet because it is probably the easiest comparison. When widespread internet adoption was introduced into everyday life, people were understandably concerned about it. I vaguely remember some news stories about this from my childhood. The memories are foggy and sometimes I wonder if I am hallucinating (see what I did there) them but I would swear the memories are real. People had concerns about the internet. Some of those concerns turned out to be completely justified. The internet gave us scams, identity theft, cybercrime, misinformation, harassment, exploitation, and entirely new ways for people to be terrible to one another, and all of those things still exist. Yet I doubt many people would seriously argue that the internet has been a net negative for society.

Think about how deeply integrated it has become into everyday life. Banking, education, communication, entertainment, medicine, commerce, transportation, research, government services, publishing, and countless other systems now depend on it in one way or another. Could we survive without it? Of course we could I would never be so intellectually dishonest to try and argue otherwise. Humanity managed for quite a while without the internet so it is not essential. However, would modern life become significantly more difficult without it? Absolutely. That does not mean every development brought about by the internet was good. It means the technology became useful enough that society adapted around it while continuing to wrestle with its problems. I suspect AI is going to follow a very similar path.

As an author, yes, I use AI. I use it for research, organizing notes and thoughts, troubleshooting, and helping me build Python programs for projects I use myself. I use it when I am trying to teach myself something new, whether that is Python, constructed languages, nutrition, computer troubleshooting, or even how to manage my social media because that is one area I have never been particularly good at. What I do not do is type, “Write me a novel,” copy the result, slap my name on it, and call myself an author. There is a difference, and I think that difference gets lost in a lot of conversations surrounding AI.

Using a tool to assist you is not necessarily the same thing as asking that tool to replace you. If I use a calculator while checking numbers in a manuscript, I still wrote the book. If I use spellcheck, I still wrote the sentence. If I hire an editor who tells me a chapter is terrible and needs to be rewritten, I still have to rewrite it. If I use software to organize two hundred pages of worldbuilding notes so I can find something six months later, the software did not create the world. The better question, at least for me, is not simply whether AI was used, but what it was used for.

I recently saw a TikTok from a creator using the handle Wild Mother who said something that stuck with me: “What fascinates me is people only seem to object to assistance when ordinary people gain access to it.” She went on to compare AI assistance with forms of support that have existed for a very long time. Have enough money to hire researchers? Great. Hire an editor? Professional. Hire someone to build your website, organize your schedule, manage your advertising, proofread your work, analyze data, or teach you a new skill? Nobody thinks twice about it.

I think there is something worth examining there. Historically, having access to expertise has often depended on having access to money. A business owner with enough capital can hire accountants, programmers, graphic designers, consultants, marketing teams, assistants, researchers, and lawyers. Someone without that kind of money has traditionally had two choices: learn everything themselves or go without. AI changes this choice because it allows someone who does not know how to do certain things figure it out. It can help someone understand terminology they have never encountered, or walks a person through a complicated process when there is nobody else around to ask. It can provide a starting point for learning something that previously seemed completely inaccessible. That does not make the user an expert, but it can lower the barrier to becoming more capable.

This is also where some of my frustration with the AI debate starts to come in, because a lot of people arguing most aggressively about this technology do not seem to understand how it actually works. I do not mean that as an insult to anyone who simply does not know. Most people do not need to understand the inner workings of a large language model any more than they need to understand the electrical engineering inside their phone. I certainly do not understand every technical aspect of it myself. The problem comes when someone has a vague conceptual idea of what the technology does, treats that idea as fact, and then builds an entire moral argument on top of it.

You still hear people describe generative AI as though there is a giant database somewhere containing every book, painting, photograph, and article ever created, and when you enter a prompt the system simply finds pieces of those works and stitches them together. That is not an accurate description of how modern generative models work. Large language models learn statistical relationships within enormous amounts of training data and use those learned patterns to generate new text. Image-generation systems work differently, but the misconception is similar: they are not normally searching through a giant folder of existing images and assembling a collage out of whatever they find.

That does not mean the copyright and training-data debate disappears, and it certainly does not mean models are incapable of reproducing material they encountered during training. Those are legitimate issues. But saying, “The machine looked at copyrighted material, therefore every output is a stolen copy,” skips over an enormous amount of technical reality.

The same problem happens with hallucinations. AI models can hallucinate. Once again I am not so intellectually dishonest to argue that hallucinations never happen. AI can give you a confident answer that is completely wrong, invent a source that does not exist, misunderstand context, or make an assumption that was never actually supported. Anyone using these systems for factual work should understand that and verify important information and I do understand this. What frustrates me is the tendency to talk about hallucination as though the technology has been frozen in time and has not improved.

The fact that an AI model can be wrong is also not the devastating criticism some people seem to think it is. Humans are wrong all the time. I mean, how many times have experts been wrong? I remember such lovely assurances as ‘too big to fail,’ or the repeated claims that Iraq possessed weapons of mass destruction to name two examples. Or we could bring up classical search engines returning bad information. Some of you may be too young to remember early Google, but I am not, and I can tell you it was not great. Books contain errors. Journalists issue corrections. Experts disagree with one another. The relevant question is not whether an information tool can ever be wrong. The relevant questions are how often it is wrong, how serious those errors are, whether they can be detected, and whether the person using the tool understands its limitations.

What bothers me most is watching someone repeat a criticism they heard in a thirty-second TikTok as though it settles an enormously complicated technical and ethical discussion. If you want to oppose AI, oppose it. There are plenty of intelligent arguments to make about copyright, labor, consent, environmental cost, privacy, and corporate power. Make those arguments. But if your entire understanding of the technology is “it steals pictures” or “AI just makes things up,” then perhaps spend some time learning how the systems actually function before announcing that everyone who uses them is unethical.

I am not saying you need a computer-science degree before you are allowed to have an opinion. I certainly do not have one. I am saying that if you are going to make sweeping claims about what a technology is doing, especially claims serious enough to accuse other people of theft or dishonesty, then there should be some responsibility to understand the thing you are criticizing.

And that brings me to the argument that probably irritates me more than any other in the AI debate: the blanket statement that “AI is theft.”

It gets repeated constantly as though the issue has already been settled and there is nothing left to discuss. The problem is that the reality is nowhere near that simple. We need to stop pretending that all training data is the same. Modern AI companies use mixtures of publicly available information, licensed data, data obtained through partnerships, material generated by human trainers and researchers, user-provided data where permitted, synthetic data, and other sources. That does not mean every company has handled every dataset perfectly, and it does not mean there have not been serious disputes over copyrighted material. There absolutely have been. But turning that complicated landscape into “AI stole everything” is not an accurate description of how these systems are built.

There is also an important difference between publicly available and public domain, because they are not the same thing. Something being available on the internet does not automatically strip it of copyright. That is part of what makes the legal and ethical argument surrounding AI training complicated in the first place, and it is exactly why I dislike seeing such a complicated subject reduced to a slogan.

The word “theft” starts doing more emotional work than analytical work at that point. Training a model on information is not the same process as copying a book, removing the author's name, and selling that book as your own. A language model does not normally contain a searchable library of complete works that it retrieves whenever you ask it a question. It learns statistical relationships between language, concepts, structure, and context and uses those learned relationships to generate new outputs. That does not automatically answer the copyright question, but it absolutely matters when discussing what the technology is actually doing.

There are legitimate concerns about memorization. There are situations where models have reproduced portions of material encountered during training, and companies should be expected to reduce that behavior. There are legitimate questions about whether certain copyrighted works should have been included in training datasets at all, whether creators should have had an opportunity to opt out, whether licensing should be required in certain circumstances, and how compensation might work. Those are real arguments, and pretending they do not exist would be just as dishonest as pretending every AI output is stolen.

There have also been companies and datasets that deserve criticism. If someone obtains material illegally, ignores licensing requirements, bypasses access controls, or builds a system designed to reproduce protected works on demand, then criticize that behavior. Hold that company responsible for what it actually did. Regulate it if regulation is needed. Change the law if the law needs to change.

What I reject is the jump from “some training practices may be unethical or infringing” to “therefore generative AI as a technology is theft.” That logic would be absurd almost anywhere else. If a photographer steals an image, photography is not theft. If a programmer pirates software, programming is not theft. If someone plagiarizes a novel, word processors are not theft. If a company illegally collects personal information, databases are not theft. We identify the improper behavior and the person or organization responsible for it. We do not declare the underlying technology morally contaminated forever because somebody used it badly.

AI should be treated the same way.

If a model was trained using improperly acquired copyrighted material, deal with that specific model, company, or dataset. If a company licensed its data, created its own data, used public-domain material, gathered publicly accessible information in a way the law permits, or generated synthetic training data, those circumstances are different and should be treated differently. That nuance matters.

There is another uncomfortable part of this conversation as well. Human creativity has never occurred in isolation. Writers learn by reading other writers. Painters study other painters. Musicians absorb thousands of songs before developing their own sound. Nobody sits down to create something with an entirely empty mind. We spend our entire lives absorbing patterns, structures, techniques, ideas, vocabulary, imagery, and experiences from other people's work.

I want to be careful with that comparison because a human brain and a machine-learning model are obviously not the same thing, and pretending they are would be once again intellectually dishonest at best. The legal rights attached to a person's work also do not disappear simply because someone else can learn from it. But the comparison is useful for one very specific reason: learning from something and reproducing that thing are not automatically the same act.

If an AI spits out a near-verbatim copy of someone's novel and another person sells it, I have absolutely no problem calling that wrong. If someone prompts an image generator with the explicit goal of perfectly imitating a specific artist and then pretends that artist created the result, I understand why people object. If someone's voice is cloned without their consent and used commercially, I think they should have protections. Those situations do not, however, describe every use of generative AI any more than plagiarism describes every person who owns Microsoft Word.

So when someone tells me “AI is theft,” my first question is increasingly becoming: what exactly are you accusing of being stolen? Are we talking about the training data, the output, the style, the author's labor, someone's likeness, someone's voice, or an economic opportunity? Those are different accusations, and they require different arguments.

If we are going to have this debate, then let's have the actual debate. Let's talk about licensing, fair use, consent, opt-outs, memorization, compensation, transparency, and what rights creators should retain when their work is publicly accessible online. Those are difficult questions, and I do not pretend to know exactly where every legal or ethical boundary should ultimately land. “AI is theft” does not answer any of them. It just ends the conversation before it starts.

All of this is exactly why I think there are legitimate ethical lines surrounding AI and why those lines should be discussed seriously. Consent matters. If someone's voice, artwork, writing, likeness, or other creative work is being replicated in a way that directly appropriates their identity or work, the person behind that work deserves consideration and, depending on the circumstances, protection. Passing machine-generated work off as something you personally created raises obvious ethical questions. Using AI to deliberately deceive people, manufacture evidence, impersonate someone, plagiarize, scam, or spread misinformation is wrong.

But once again, those are arguments about how a tool is being used. They are not automatically arguments against the existence of the tool itself.

The environmental argument is another area where I think the conversation often loses perspective. This one frustrates me, not because environmental concerns surrounding data centers are imaginary. They are not. Large-scale computing consumes electricity, requires infrastructure, and in many cases uses significant amounts of water for cooling. Those impacts deserve scrutiny, and companies building these systems should absolutely be pushed toward more efficient and sustainable practices. What bothers me is when AI is discussed as though it exists in an environmental vacuum.

Modern society is filled with industries and conveniences that consume staggering amounts of electricity, land, fuel, and water. Data centers existed long before the current AI boom because streaming video, cloud storage, social media, online shopping, banking, search engines, gaming, and the rest of the modern internet all require physical infrastructure. Golf courses consume water. Lawns consume water. Streaming movies requires data centers. Storing millions of photos in the cloud requires data centers. Watching hours of social media requires servers somewhere to deliver those videos. None of that excuses AI's environmental cost, but it does mean we should be consistent about how we discuss environmental cost.

If sustainability is the concern, then let's talk seriously about sustainability. Let's talk about energy sources, cooling technologies, water usage, efficiency, infrastructure, and where these facilities are being built. Reducing the entire discussion to “AI uses water, therefore using AI is immoral” is not a particularly useful environmental argument, especially when so much of modern life relies on resource-intensive infrastructure we rarely stop to think about.

The same applies to employment. Yes, AI will change jobs, and it already is. Some roles will probably disappear, others will change dramatically, and new jobs will appear that did not previously exist. That uncertainty is uncomfortable, especially for people whose livelihoods may be directly affected by the transition, and that deserves empathy and serious policy discussion.

But technological change disrupting employment is not new. Automation transformed manufacturing. Computers transformed office work. The internet transformed retail, journalism, advertising, travel agencies, publishing, entertainment, and countless other industries. Digital photography devastated businesses built around film. Streaming transformed video rental. Smartphones replaced dozens of standalone products. We did not stop developing those technologies. We adapted, sometimes badly and sometimes successfully, and I suspect we are going to have to adapt again.

None of this means you have to like AI or use it. If you have ethical concerns that make you uncomfortable incorporating it into your writing, your job, your business, or anything else, I completely respect that. You can believe the way certain models were trained was unethical. You can believe the law has not caught up with the technology. You can believe companies have handled artists, writers, workers, or data poorly. Those are all arguments worth having.

What I struggle with is the leap from “I do not want to use this” to “nobody who uses this can be ethical.” Those are two very different positions.

There are thoughtful ways to use AI and irresponsible ways to use it. There are applications I find genuinely exciting and others that make me deeply uncomfortable. There are companies doing things with this technology that deserve criticism. There are questions about regulation, ownership, transparency, labor, and consent that society has not adequately answered yet. I do not think pretending otherwise helps anyone. At the same time, I also do not believe AI is disappearing.

The technology is already being incorporated into software, search engines, phones, operating systems, businesses, medical research, cybersecurity, accessibility tools, education, programming, logistics, and countless other areas. Many people will interact with AI-powered systems without consciously sitting down and deciding that they are “using AI,” and that integration is only going to make the line increasingly difficult to draw.

Maybe, then, the conversation needs to move beyond whether AI itself is inherently good or evil and focus more on responsibility. Who created the work? Was consent involved? Was someone deceived? Was copyrighted material reproduced? Was a human making the final decisions? Was the tool assisting someone's ability or replacing something they were claiming as their own? Was the technology being used to educate, organize, create, manipulate, impersonate, exploit, or deceive?

I am going to continue using AI. I am going to use it to learn, research, organize information, troubleshoot problems, challenge ideas, and help me work through things I do not yet understand. I am also going to remain the person writing my books. Those two things are not incompatible.

AI does not have to replace human creativity to be useful. It does not have to replace expertise to make expertise more accessible. It does not have to do everything for us to help us do certain things better. Like every other powerful tool, what happens next is going to depend largely on what we decide to do with it.

Love it, hate it, use it, or do not use it. Criticize it when criticism is deserved. Demand better laws, better transparency, better environmental practices, and better protections when those things are needed. Just do not mistake refusing to use a tool for proof that everyone who does use it is doing something wrong.

And if you are going to criticize it, criticize what it actually is, what it actually does, and the people or companies actually doing something wrong. Do not build your argument around a caricature of the technology simply because that version is easier to hate.

And if you’re going to criticize me for my use of it, or for this blog post, go ahead. I honestly don’t care. People will agree and disagree with me just like they agree and disagree with everything else. Everyone is entitled to their opinion. Just don’t mistake your opinion for irrefutable fact, and don’t mistake your personal moral position for universal moral authority.

Because whether any of us particularly like it or not, I have a feeling AI is going to be around for a while.