AI. It's my turn to talk about it.

🚀 We're in the Future

First I have to point out how cool of a world we now live in. Generative AI may seem normal by now, almost mundane. But when I step back and try to look at it from a fresh perspective, it's the most spectacular and unexpected technology of my lifetime. In early 2022, if you had showed me Claude Sonnet 4 iteratively implementing a feature while writing tests in Cursor and asked me to predict when such technology would be invented, I would have guessed the 2040s at the earliest. Yet here we are just a few years later in 2025.

A lot of Sci-Fi AIs suddenly look plausible with current technology. HAL 9000 from 2001: A Space Odyssey, the terminal AIs in Book of the Long Sun, Auntie Dot from Halo: Reach, to name a few. These AIs are all flawed and are not AGI (as OpenAI defines it at least) but they are able to understand what is said to them, speak intelligently, synthesize information, and make decisive decisions in general contexts, not unlike our quaint little ChatGPT 4o can today.

🧮 The Accelerationists and the Luddites

Where does it go from here? Visions are polarized, especially in the industries generative AI is being adopted in. Some get hyped about the impending singularity whenever a new model is released. Others shrug it off as being of little consequence.

I suspect many in both camps are coming from places of fear, but with different, almost opposite sources. The "Luddites" are afraid of being replaced by AI and marginalized as a result. While many of the Accelerationists already feel marginalized, or are haunted by an 'is this all there is?' kind of ennui, or fear humanity as a whole is careening towards a cliff. For them, the tantalizing Sci-Fi future that AI seems to herald is the cure.

When we see viewpoints get so polarized, historically that seems to be a great hint about how it's really going to unfold. Ultimately the truth is going to lie somewhere in between the two poles, likely pretty close to the center. The software industry will be disrupted (I think this is irrefutable by this point). But are we going to be uploading our minds to computers and lording over our own star systems next year? Probably not.

Generative AI will continue to disrupt industries. We'll get better and better at applying it. But as we do so its limitations will become more and more clear, leaving us far short of any utopian (or dystopian) vision.

This doesn't preclude those possibilities eventually, as technology only improves, but rather than a sudden radical upheaval we're more likely to experience years that don't look much different from the previous, despite significant changes on the decade timescale. This is how it's always been.

💀 My Death as an Artist

Outsiders imagine programming as a pretty rote and mechanical task. Someone decides what the computer should do, and the programmer sits down and "types it in". But if you're reading this, you know that's not how it works.

We code with our own voices, and no two programmers make all of the same decisions. We get better at it (and sometimes worse) over time, but there's no universal ideal that we converge on. We remain unique, like painters or authors, and we take pride in our work.

But when AI writes code, I don't feel like it's mine. Sure, I'll make tweaks by hand and prompt the AI to make changes, but I don't rewrite it into my own voice, there would be no point using AI then.

So I find that I don't care about code that AI generates for me. I still care about the architectural choices and the functionality, but I become detached from the microscopic details of the code and any artistry that might have been there. And I worry this detachment might hurt the product itself, that's something I’m still trying to navigate effectively.

This is why I haven't been using AI much for personal projects. I use Copilot autocomplete, but I rarely prompt AIs to write large pieces of code for me. I might be sacrificing efficiency, but I find it more fun this way. I get into the flow and care about the code.

🦋 The Struggle as Metamorphosis

What's going to happen to beginner programmers in this age of AI? There's a lot of trepidation about this, and it's well-deserved. I'm not talking about the idea of AI replacing programmers, but rather how beginner programmers will (or won't) emerge from their cocoons one day as expert programmers.

I vividly remember the first program I ever wrote. It was a script for Garry's Mod in 2006. It was about 50 lines long. It was attached to an AK-47, and if you picked up that AK-47, it froze you in place until you admitted in chat to being a n00b. It was trivial, but it took me about a week of banging my head on it late into the night struggling to interpret Lua documentation like ancient hieroglyphs. For that time, I was obsessed with it and thought about little else. When I was done, I knew the basics about variables, functions, and events.

But now? ChatGPT could spit that out for me in 30 seconds. Nobody needs to struggle to build something like that anymore. I could have it working in no time, but I'd hardly learn a thing from the experience.

The fact that AI is capable of this doesn't just deprive beginners of valuable learning experiences. That part has a clear, if painful, solution for anyone disciplined enough (just resist the temptation and don't use AI during your formative stages).

No. More insidiously, it's also deeply demoralizing.

For as long as I can remember, I have dreamt of being a respected electronic music producer. It's my secret life goal. I've never dedicated the time to it that a life goal deserves, but I have dabbled with it and know basic DAW (Digital Audio Workstation) usage.

Until recently, this basic knowledge made me a better electronic music producer than the 99.9% of humanity that has never touched a DAW before. But that's no longer the case. I now have thousands of hours to go until I'm half as good as the AI music generators that everyone now has access to. Realizing this, I now struggle more than ever to put in effort. Beginner programmers are experiencing the same thing.

Where does this lead economically? The world still needs programmers with deep technical expertise, but the pipeline from beginner to expert is being squeezed. On one extreme, it's possible that AI will win the race. That is, AI replaces all or almost all programmers in the 30-40 years between now and when the current generation of programmers retires. 30-40 years is a long time, after all. On the other extreme, maybe we end up in a Dune-like world largely starved of the talent still needed to maintain the machines built by previous generations. As usual, the reality is likely somewhere in between, perhaps with "vibe coding" as the main development style, supported by a caste of (perhaps mostly older) programmers who uniquely retain the ability to go deeper.

And where does it lead us as artists? I fear that, like the act of using AI to write code that's not my own, the watering down of the learning experience may be another factor that could drain life from the experience of programming, especially for those too young to have experienced it.

☁️ The Realization of the Dream

There's another interpretation we can take of my experience with music. Sure, I might now be farther than ever from micromanaging those sine waves and drum loops into a unique and personal thing of beauty. And yet paradoxically, I'm closer than ever to making something. I can, and have, dreamt something up, asked an AI to build it for me, and received a pretty good result. This is true of music and digital art as well (including the cover image of this very article). No, it doesn't feel like it's truly mine, nor is it exactly what I had envisioned, but at least it's something - something I'm currently incapable of building otherwise.

Moreover, I expect this experience to become more personal over time, not less, as the tools will come with more customizability to help us achieve exactly the sound/look that we had imagined.

I expect this will happen with programming as well. The tools will do better at doing what we ask and realizing our dreams exactly. For me as an experienced programmer, that makes it feel less personal (since I'm not writing the code myself). But for a non-programmer, it's more personal (because they can get exactly what they envisioned without compromise).

This is the bright side of applying AI to any art (including programming). It can help anyone become a builder, and realize dreams that may otherwise have been out of reach, or at least locked behind thousands of hours of study and practice.

Conclusion

I do have some more thoughts but I like that this is ending on a positive note, and it's long enough already, so I will wrap it up here.

The impacts of AI on our society are already far-reaching and will become even more so. We'll all be winners in some ways and losers in others. I am struggling with a sense of loss in the AI age, but also optimistic about the opportunities it comes with. We're not just changing how we build things, but what we can build, and that's exciting.

What do you think?