AI IN MI I BY PETER DODS The Anti-AI Argument
A cross several articles, I’ve been an unapologetic AI cheer- leader. I’ll keep showing what these tools can do. But a growing number of people are pushing back, and much of their criticism deserves a fair hearing. Here are a few “devil’s advocate” arguments against AI that I think everyone should be aware of. PRESSURE ON INFRASTRUCTURE The infrastructure powering this technology drives real resource
On morality, outputs have con- sequences: who gets the loan, what shows up in the feed, which resume gets filtered. “It’s just math” easily becomes a way to shirk responsibil- ity. If you rely on an LLM’s output without verifying it, you’re doing it wrong. THE BUBBLE It’s hard to see $20 monthly sub- scribers converting into $1,000 power users fast enough to justify trillions in infrastructure com- mitments. The only viable path to those massive returns is enter- prise software. Can any provider secure durable pricing power be- fore the tech stack commoditizes? If not, the unwind is harder to predict than the dot-com bust. FORWARD-THINKING FASTIDIOUSNESS Much of this comes back to job displacement. Beyond that, the tech itself cuts both ways. Will Suno Studio, a full AI music produc- tion suite, grow the music-making population or decimate it? I got into this industry as an Ableton beat maker. That hobby led me to buying a failing music store 20 years ago, and then falling in love with electric guitars. Would my 25-year-old self have bothered learning to chop samples if he could just type a prompt and get a finished song? Or does this lower barrier flood the market with new enthu- siasts, the way Serato did for DJs? It’s hard to say. What I know is the customer walking into our stores five years from now will have been fundamen- tally shaped by AI. MI
strain. Nationally, U.S. data centers use a fraction of a percent of America’s freshwater, far behind what golf courses consume. But the national number hides the local story: A facility that barely registers on a chart can still squeeze a drought-prone region like Arizona. The grid faces a similar crunch. Data centers used roughly 183 TWh in 2024, which is more than 4% of domestic electricity, with demand projected to rise sharply through 2030. A bright spot for both pressures is on-device AI. I can run Gemma 4 E4B on my phone, and for everyday tasks it holds its own against the GPT-3.5 we met in 2022. COGNITIVE OFFLOADING One warning sign of this is using AI in a way that weakens a skill you already have. Coding was never my strength, so leaning on AI for JavaScript isn’t at- rophying my brain. What I won’t do is hand AI my final decisions. I’ll ask for pros and cons, but the call is mine. JOB DISPLACEMENT This is a prisoner’s dilemma. Take the luddite path and you probably won’t slow the technology down.
A few criticisms of AI that deserve a fair hearing.
Or you can embrace it and grab as much productivity from it as you can. History says new technology creates new jobs, but this shift feels different. For now, many AI layoffs look overblown. INTELLECTUAL PROPERTY THEFT These models are trained on vast amounts of copyrighted material, and the lawsuits are still moving through the courts. AI CAN LIE & IT ISN’T MORAL I’d push back on both. A meteorologist who calls for sun and gets rain isn’t a liar. An LLM producing the most probable next token isn’t ly- ing when it’s wrong. It’s a prediction machine. It doesn’t know the truth, and it doesn’t have values. The fairer critique is that AI sounds confident whether it’s right or wrong.
Peter Dods is the owner of Honolulu- based Easy Music Center.
26 I MUSIC INC. I SEPTEMBER 2026
Powered by FlippingBook