{"id":737,"date":"2026-04-13T20:35:12","date_gmt":"2026-04-13T20:35:12","guid":{"rendered":"https:\/\/millermattson.com\/dave\/?p=737"},"modified":"2026-04-13T23:28:08","modified_gmt":"2026-04-13T23:28:08","slug":"the-machine-that-stole-my-hobby","status":"publish","type":"post","link":"https:\/\/millermattson.com\/dave\/?p=737","title":{"rendered":"The Machine That Stole My Hobby"},"content":{"rendered":"\n

For a couple of decades, I spent happy hours making electronic music<\/a>. Not professionally \u2014 I never played a gig, never really expected anyone outside my immediate circle to listen. That wasn’t the point. The point was sitting in front of a screen with a cup of coffee, nudging a bassline until it felt right, layering in some texture, being surprised when a combination of sounds suggested a mood I hadn’t anticipated. I was, by any objective measure, a mediocre amateur composer. But I was a mediocre composer who was genuinely, consistently having fun.<\/p>\n\n\n\n

I also used to enjoy building small software projects<\/a>. Not professionally. Just for myself \u2014 a little tool to organize something, a toy program to explore some idea. The pleasure was in the whole arc of it: starting with a vague intention, sketching out how the pieces might fit together, writing the actual code, running into problems I hadn’t anticipated, debugging for an infuriating hour only to find a single misplaced character, and then finally \u2014 finally<\/em> \u2014 watching the thing work. The software was usually modest and often imperfect. But it was mine, and I had made it, and that meant something.<\/p>\n\n\n\n

I don’t do either of these things much anymore. Not because I’ve run out of time, or ideas, or interest in the subject matter. I’ve stopped, or nearly stopped, because AI has made both activities feel strangely pointless. A current AI music tool can generate a more polished, more sophisticated track than anything I could produce, in minutes. An AI coding assistant can take a project I might have spent two or three evenings on and produce a working version just as fast. The results are often genuinely impressive. And they have effectively drained the fun out of the hobby.<\/p>\n\n\n\n

I’ve been trying to figure out why, and whether this feeling is something new, or just the latest version of a complaint that every generation makes about every new technology.<\/p>\n\n\n\n

The fun, I think, was spread through the whole process, not concentrated at the end in some finished thing. When I was making music, having an idea was fun. Trying to realize it and failing was frustrating in a way that was somehow also fun. Learning what a particular tool could and couldn’t do, finding workarounds, discovering that a mistake had produced something more interesting than what I’d originally intended: all of this was engaging in the way that good puzzles are engaging. The song at the end was partly a trophy and partly beside the point. What I had really gained was time spent in a state of absorbed attention, plus a slightly expanded set of skills I hadn’t had before. I had become, incrementally, someone who knew a little more about making music than I had known last week.<\/p>\n\n\n\n

The same was true of software. The process felt satisfying: concept, architecture, implementation, debugging, completion. Each stage had its own texture, its own small difficulties, its own minor triumphs. The finished app was real and useful, but the more durable reward was the feeling of having navigated the whole journey. The confidence that I could do it again.<\/p>\n\n\n\n

A hobby is not a production pipeline. It’s a space where effort is allowed to be the point, where the difficulty isn’t an obstacle to the goal but is, in a lot of ways, the goal itself.<\/p>\n\n\n\n

When AI turns “I have an idea” into “here it is” almost instantly, that space collapses. The process disappears. The incremental learning disappears. The mistakes \u2014 which, in hindsight, were doing a lot more work than I appreciated \u2014 don’t really occur, or occur so briefly and manageably that they leave no impression. You are no longer a maker working through a problem. The result gets better, but your role in it gets thinner.<\/p>\n\n\n\n

That’s probably why an impressive AI-generated result can feel emotionally deflating rather than exciting. If you cared mainly about having a good song or a working app, you should be thrilled \u2014 you have one, and it took almost no time. But if the deeper reward was becoming the person who could make the song or the app, then the AI’s efficiency is working against you. It has handed you the trophy without the race.<\/p>\n\n\n\n

Professionals experience AI differently, and it’s worth considering why. For a working musician or a software engineer, AI tools can mean speed, scale, and the ability to take on more ambitious projects. The process has always been partly a means to an end for them. AI assistance can feel like an upgrade. But for amateurs, the process was never instrumental. It was the whole reward. Which is why the same tool that feels like a gift to a professional can feel like a kind of theft to a hobbyist.<\/p>\n\n\n\n

\n\n\n\n

Of course, maybe this is just the standard complaint. Every generation has told itself some version of the same story: the latest technology has finally gone too far, has finally replaced something irreplaceable, and things were better before.<\/p>\n\n\n\n

There are good historical analogies, and the most interesting one is hiding in my own example. Electronic music \u2014 the very genre I was composing in \u2014 was itself built on a sequence of automation tools that each provoked similar anxieties. Synthesizers removed the need to use acoustic instruments. Drum machines replaced session drummers. Sequencers meant you didn’t have to play a part in real time to have it appear in a recording. Digital audio workstations put professional-grade production in the hands of amateurs in their bedrooms. At each step, someone complained that something essential was being lost, that the new tool was making music too easy, that the result was somehow less legitimate.<\/p>\n\n\n\n

And yet musicians adapted. The tools didn’t kill music-making; they changed what music-making meant, and shifted which skills were valued. A bedroom producer with a laptop developed real craft \u2014 in arrangement, in sound design, in mixing \u2014 even if that craft looked quite different from what a trained instrumentalist had developed. The technology opened doors for people who might never have made music otherwise.<\/p>\n\n\n\n

But there’s a difference. Each of those earlier tools automated execution. They made it easier to produce the sounds you had in mind while leaving the compositional decisions to you. The composer still had to compose. AI feels different to me, because it can generate the compositional decisions themselves. It doesn’t just hand you a better brush; it offers to paint the picture.<\/p>\n\n\n\n

The comparison to industrialization is useful here too, in a different way. The rise of factory production in the nineteenth century didn’t eliminate craft hobbies; it transformed them. Woodworking, weaving, pottery survived and in some cases flourished, but their meaning shifted. They were no longer the primary means of producing objects; they became something people chose for their own sake, precisely because the effort was involved. The difficulty became the feature. What was once simply how things got made became a deliberate, valued practice, something set apart from the efficiency of industrial production.<\/p>\n\n\n\n

Something similar may be happening now, just faster and more comprehensively.<\/p>\n\n\n\n

There’s an obvious response to all of this: AI is lowering barriers. People who previously couldn’t make music because they lacked technical skills can now make music. People who wanted to build software but didn’t know how to code can now build software. Isn’t that a good thing? Isn’t the democratization of creative expression worth some disruption to the hobbies of people who were already doing it?<\/p>\n\n\n\n

Yes, probably. I don’t want to dismiss this. There is something genuinely valuable about tools that make creative expression accessible to more people, and I’m aware that my attachment to the older, harder version of the hobby has a self-serving quality. Romanticizing difficulty is easy when you’ve already gotten past the difficulty.<\/p>\n\n\n\n

But there’s a real distinction between accessibility and satisfaction. A tool that makes something easier to start doesn’t automatically make it more rewarding to pursue. If the struggle was genuinely part of what made the hobby meaningful, removing it doesn’t simply lower the floor. It may also lower the ceiling. The music is easier to start. It is also, perhaps, less yours when it’s finished.<\/p>\n\n\n\n

\n\n\n\n

I haven’t fully worked out what to do about any of this. But I’ve started to wonder whether I’ve been framing the problem too narrowly.<\/p>\n\n\n\n

The anxiety I’ve been describing assumes a particular model: AI as a replacement, stepping in and doing the thing for<\/em> you, leaving you with a result and nothing else. And that model is real \u2014 it’s what happens when you hit a button and receive a finished song, or paste a prompt and get a working app.<\/p>\n\n\n\n

But there’s another model, one that feels meaningfully different: working with<\/em> AI as a collaborator you direct. Not delegating the creative act, but using AI as a capable assistant who executes within a vision that remains yours. The decisions \u2014 the goals, the structure, the tone, the choices about what to keep and what to revise \u2014 stay with you. The AI handles execution you’d otherwise struggle with or labor over. What you’re left with isn’t a trophy handed to you; it’s something closer to a conductor’s relationship with an orchestra. The musicians play the notes. But the interpretation, the phrasing, the overall shape of the thing \u2014 that’s yours.<\/p>\n\n\n\n

I’m not sure this completely resolves the loss I’ve been describing. Some of what I miss is genuinely gone: the particular texture of debugging alone at midnight, the specific satisfaction of a bassline I nudged into place myself. Those experiences were real, and I don’t think collaboration fully substitutes for them.<\/p>\n\n\n\n

But I’ve come to think the picture is not necessarily bleak. The difficulty that made those hobbies meaningful was always, at its core, the difficulty of making decisions<\/em> \u2014 of figuring out what you wanted and finding a way to get there. AI can assist with execution without necessarily taking that away. Whether it actually does depends on how you use it, and perhaps on the deliberate choices you make about where to stay in charge.<\/p>\n\n\n\n

Something may have changed. But perhaps not everything worth preserving has been lost.<\/p>\n\n\n\n

\n\n\n\n

Epilogue and full disclosure<\/h2>\n\n\n\n

The essay above was written mostly by AI. This epilogue, however, is entirely mine, written with no AI help.<\/p>\n\n\n\n

I invested several hours in collaboration with multiple AI agents to produce the essay. I created the original thesis, then acted as editor and advisor with multiple LLMs. It required multiple revisions to get the outline just right, then more revisions to refine the opening, conclusion, and internal structure. It took several revisions to refine the vocabulary and phrasing. As a final step, I manually changed only a few words and phrases.<\/p>\n\n\n\n

This is an example where an AI produced a finished product that is better than what I could have written. Yet, it was a collaborative effort and did not entirely deprive me of a sense of creative accomplishment. The detailed wording is not mine, but the message and tone is all mine. I felt in control, and I felt supported by a talented assistant.<\/p>\n\n\n\n

Machines are talented, but they don’t feel. They have no greed, no aspirations, no fears. Until they achieve a human-like consciousness — which is unlikely to happen anytime soon — humans will be the conductors of mechanical talent. And there’s reward in that.<\/p>\n\n\n\n

In some way, I hold onto this as a hopeful future with AI.<\/p>\n\n\n\n

<\/p>\n","protected":false},"excerpt":{"rendered":"

For a couple of decades, I spent happy hours making electronic music. Not professionally \u2014 I never played a gig, never really expected anyone outside my immediate circle to listen. That wasn’t the point. The point was sitting in front of a screen with a cup of coffee, nudging a bassline until it felt right, […]<\/p>\n","protected":false},"author":2,"featured_media":740,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[83,130,131,110],"tags":[86,129,104,128,109],"class_list":["post-737","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai","category-hobbies","category-programming","category-writing","tag-ai","tag-hobbies","tag-llm","tag-software-development","tag-writing"],"_links":{"self":[{"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=\/wp\/v2\/posts\/737","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=737"}],"version-history":[{"count":4,"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=\/wp\/v2\/posts\/737\/revisions"}],"predecessor-version":[{"id":744,"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=\/wp\/v2\/posts\/737\/revisions\/744"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=\/wp\/v2\/media\/740"}],"wp:attachment":[{"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=737"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=737"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/millermattson.com\/dave\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=737"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}