I love AI.
I love what it lets me do in software development, in design, in writing. It has opened doors I genuinely never thought I’d walk through. That’s not a talking point — it’s just true, and I’m done softening it to make it more palatable.
I know some of you hate it. I know that hate runs deep. And I don’t think you’re wrong to feel it.
But I think you’re feeling something humans have felt many times before. And I think it’s worth understanding what that pattern actually looks like — because the feeling is usually more right than the conclusion.
John Philip Sousa Was Not an Idiot
In 1906, John Philip Sousa — “The March King,” one of the most beloved and respected musicians in America — published an essay called “The Menace of Mechanical Music.” He warned that phonographs and player pianos would replace human skill, intelligence, and soul with mechanical reproduction. That music would lose its humanity. That the art form itself was under existential threat.
He was wrong about almost all of it. Recorded music went on to birth entirely new art forms — jazz, blues, rock, hip-hop, electronic music. It inspired generations of people to become musicians. The soul didn’t go anywhere.
But here’s what usually gets left out: Sousa wasn’t just being hysterical. He had a real, legitimate grievance underneath the moral panic. He was fighting for composer royalties. He testified before Congress and helped bring about the Copyright Act of 1909, which created the mechanical royalty system that still protects musicians today.
His economic argument was sound. His apocalyptic argument about the death of musical soul was not.
That distinction is the whole thing.
This Script Is Very Old
When Gutenberg’s press spread across Europe in the 1450s, a German abbot named Johannes Trithemius wrote a passionate defense of hand-copying manuscripts. Scribes who stopped copying would grow spiritually lazy, he argued. Something sacred would be lost. The craft would die and take something irreplaceable with it.
He had his treatise printed on a press so it could reach a wider audience.
When the daguerreotype was announced in 1839, the painter Paul Delaroche reportedly declared “From today, painting is dead.” The poet Charles Baudelaire called photography the refuge of failed painters — a soulless mechanical process with no imagination, no judgment, no soul. Not real art. The Metropolitan Museum didn’t acquire its first photographs until 1928. Nearly ninety years after the medium’s invention. Today no serious person questions whether photography can be art.
When desktop publishing arrived in 1985, professional typesetters who had spent careers mastering their craft watched PageMaker let anyone do their job on a desktop. When digital photography and Photoshop went mainstream, film photographers called digital editing cheating — insisting real photography meant getting it right at capture, not fixing it after. When Auto-Tune emerged in the late 1990s, critics called it a plague, a robotic fake, an anaesthetic applied to human emotion. Jay-Z released “D.O.A. (Death of Auto-Tune)” in 2009.
Auto-Tune is now a foundational creative instrument. Digital photography is just photography. Desktop publishing is just publishing.
The arguments against each of these tools followed the same script almost word for word: it’s soulless, it’s cheating, it’s going to destroy the craft and the people who practice it. And in every case, the craft survived, transformed, and in many ways flourished.
On the “Stolen Art” Argument
I want to address the training data question head-on, because it’s the one that sounds most damning and deserves a real answer.
The argument goes: AI was trained on artists’ work without their consent. It stole from them to learn. That makes everything it produces tainted.
Here’s what I’d push back on.
Every human artist who ever lived learned exactly the same way.
Michelangelo learned to paint by copying Giotto. Leonardo da Vinci entered a workshop at around age fourteen and spent years reproducing his master’s drawings and style before he was allowed anywhere near an original composition. Picasso absorbed African tribal masks and ancient Iberian sculptures and synthesized them into Cubism — work he never asked permission to learn from. The entire Renaissance apprenticeship system was built on the idea that you copy the masters until you understand what they know, and then you make it your own. This wasn’t considered theft. It was called training.
The famous line — “Good artists copy, great artists steal” — is attributed to Picasso, though the evidence linking it directly to him is thin. It may trace back to T.S. Eliot, or William Faulkner, or earlier. Whoever said it first, the sentiment holds: the whole history of art is influence absorbed, transformed, and made new. You cannot point to a single great artist who created in a vacuum, untouched by what came before.
So when people say AI stole from artists by learning from their work, I’d ask: what exactly did your favorite artist do? They looked at everything they could find. They internalized styles, techniques, and compositions that other people made. They were shaped by it. The difference with AI is scale and speed — not the fundamental act of learning from what exists.
There is a legitimate version of this argument, and I won’t dismiss it. The legal question of whether scraping images from the internet to train a model constitutes copyright infringement is genuinely unresolved — courts are working through it right now. There’s a real difference between a human artist spending years absorbing influence and a model ingesting billions of images overnight. The harm done to artists like Greg Rutkowski — whose name was used as a prompt tens of thousands of times to generate fake versions of his style without his consent — is real, personal, and wrong.
Consent and compensation matter. I support getting them right.
But “AI learned from existing art” is not, by itself, the scandal it’s presented as. That is exactly what learning is. Copyright law has never protected style, only specific expression — a human artist can spend their whole career painting in the style of someone they admire and never owe that person a cent. AI doing something functionally similar feels different, and I understand why. But “it feels different” and “it is fundamentally different” are not the same claim.
The Fear Is Sometimes Right
There’s a reason this pattern keeps repeating. It’s not because people are stupid or reactionary. It’s because the fear is sometimes correct — and that deserves to be said plainly.
When the power loom became commercially viable in early 19th-century Britain, skilled handloom weavers saw their weekly wages fall from around 21 shillings to under 9 shillings in fifteen years. Not metaphorically hurt. Actually impoverished, many of them permanently. Society eventually benefited from cheaper cloth. The specific generation of weavers who were displaced largely did not recover.
“It all works out in the long run” is cold comfort if you’re the one being worked out.
The current AI moment has a legitimate version of this. A 2024 survey by the Society of Authors found roughly a quarter of illustrators had already lost work to generative AI. Those are not hypothetical future losses. They are happening now, to real people.
So when I say the pattern of moral panic is familiar and historically unreliable, I don’t mean there’s nothing to be angry about. I mean two arguments are happening at once and they keep getting tangled together.
The first — about economics, consent, compensation, and fairness — is legitimate and worth having seriously.
The second — that the work is soulless, fake, not real art, an affront to human creativity — is the one that has lost every single time for six hundred years.
On Data Centers and the Whole Internet
Let’s talk about energy, because it comes up in the AI conversation as if it’s a knockout blow.
AI data centers consume a lot of electricity. US data centers consumed 183 terawatt-hours in 2024 — more than 4% of the country’s total consumption, roughly equivalent to the annual electricity demand of the entire nation of Pakistan. That number is growing and AI is a significant driver. I’m not going to pretend otherwise.
But the internet has always been an energy industry. Every platform you use without a second thought runs on the same infrastructure, draws from the same grid. Streaming video, social media, music platforms — all of it is running data centers around the clock, and most of it runs on fossil fuels.
Watching an hour of HD video on YouTube or Netflix produces roughly the same carbon as over a thousand ChatGPT queries. Spotify, Netflix, and YouTube collectively consume far more energy per American household than ChatGPT does. TikTok’s estimated annual carbon footprint is around 50 million tons of CO₂ equivalent — nearly seven times bigger than Meta’s — because most of its data centers run on coal and natural gas. The point isn’t to single anyone out. The point is that this is the internet. It has always worked this way.
To be fair: at the data center level, individual AI prompts are more power-hungry than streaming, because most of streaming’s energy is consumed by your television or phone rather than the server. AI’s footprint at the infrastructure level is genuinely higher, and it’s growing faster than most other sectors.
If the concern about AI’s energy use is genuine, it deserves to be applied consistently — not just to the newest and most controversial thing on the network. The honest version of this argument is about trajectory: data centers growing faster than the grid can green itself. That’s worth taking seriously, and it’s an argument for better policy, smarter infrastructure, and holding all tech companies accountable across the board.
This Isn’t Going Away
AI is not going away.
It is a trillion dollar industry with the resources of the largest companies on earth behind it. Entire national economies are racing to lead it. It is being woven into healthcare, education, infrastructure, and science at a pace that no amount of public anger is going to reverse. That changes the question worth asking.
We have been here before with this too. When social media arrived, the concerns were real — and many of them turned out to be justified. The research on what it did to teenage mental health, to political discourse, to our collective attention span, is damning in places. People were right to be worried. Activists, researchers, and parents spent years sounding the alarm.
Social media did not go away. It is more embedded in daily life than ever.
But that pressure wasn’t wasted. It produced real conversations about algorithm transparency, about data privacy, about age restrictions and mental health protections. The fight shifted from “stop this” to “shape this” — and that shift is where the actual progress happened.
That’s the conversation I want to have about AI. Not whether it exists, but how we use it, who it serves, what guardrails belong around it, and how we make sure the people it disrupts aren’t just left behind.
Yelling at individuals to stop using it isn’t going to stop it. It never has. What moves the needle is the harder, slower work of pushing the people building it to do it responsibly — and holding them accountable when they don’t.
Where I Land
I use AI. I’m not going to stop. It has made me more capable than I ever expected to be and I refuse to apologize for that.
I also believe the people building these tools have real obligations — to the artists whose work trained them, to the broader creative ecosystem, to getting consent and compensation right. I support that fight even as I use the tools.
Your anger isn’t unfounded. The disruption is real. Some of what’s happening is genuinely unresolved and worth fighting about.
But if your argument is that AI-assisted work has no soul — that using these tools makes you a fraud, or the output meaningless — you are making the same argument that has been made against every transformative creative technology in history.
In six hundred years, it has never been right yet.
We have been here before. We will be here again. The only question worth asking is whether we let the fear make us smaller, or whether we do the harder work: protect the people the disruption hurts, hold onto what actually matters, and move forward.
I know which side of that I’m on.