I’ve been playing with dalle-mini the last few weeks. Part of what makes it fun to play with are the bizarre and obtuse outputs. They reached that sweet spot between laughably bad and frighteningly perfect: they’re good enough to be understood and enjoyed, basically.

I think that incompleteness is part of what makes it so amusing to toy with these things, and conversely what will make future versions much less fun.

The “catastrophe scenario” of AI in the software engineering job market has gained quite a lot of popularity, with people pointing out that current models already “do a better job” (what does that even mean?) than many professional programmers. I’m not sure I completely buy into this.

TLDR We launched Podlettr - go check it out!

For nearly two years, me and my good friend Sérgio Fontes, an accomplished product designer, have been working on Podlettr - a great way to quickly catch up with your favourite podcasts. As the name implies, Podlettr is a letter from your podcasts. Reading is faster than listening, and with some AI magic, we convert hours worth of podcasts into beautiful, easy-to-read weekly newsletters.

We both have demanding full-time jobs and family duties, so we had to be pragmatic with the frameworks and architecture choices we made. Rails was my obvious framework of choice. Within weeks of the initial idea, we had a working prototype.

The LLM hype and AI bubble seem unfazed by 2026. If anything, it seems to be getting stronger with no end in sight. One thing I don’t see discussed often, though, are the long-term effects of LLMs on the joy of programming.

Lets assume that the current trend regarding LLM’s ever-increasing adoption holds for a long time, and we end up in a future where “programming” is assumed to be tantamount to operating the slot machine of future-Claude until it spits out the finished product your boss wants (or he’ll do it himself, doesn’t matter).

With that future in mind, I ask myself two questions: Will people still enjoy programming, and is that even programming in the first place?

I’d argue the answer is No for both.

I fear the endgame of LLM coding agents is to empower people who do not like coding, to the detriment of those who do. Even if you used to like coding, you will probably like it less over time because of AI.[1]

So we seem stuck in a kind of cycle: immensely powerful economic and political incentives push AI coding, which makes coding less enjoyable, empowering people who don’t like coding to become more visible in the software industry, driving higher adoption of AI coding, making coding less enjoyable again, and so on.

Its safe to say most people on the planet do not enjoy coding. That is obvious and not a problem. However, a small fraction of people do enjoy coding. I’m not thinking of Leetcoders or the typical FAANG software developer. I’m thinking of the demoscene, 64k competitions and coding golf, as well as more mundane programming language dojos and web framework conferences. I’m thinking of people that enjoy programming and are good at it - and are paid to use their talent.

In this AI-first future, I don’t see those people thriving as they did until now. Technical mastery and artistic brilliance seem to lose their luster in this hypothetical future of perfect answers. It becomes a kind of tree-falling-in-the-forest thing. If you rack your brain for weeks writing a ray tracer for your thermostat only to find out AI has already done it tenfold better? Will it still be fun, or will it be disheartening? Will it even make sense to have a code golf competition if all the winning entries are from AI bots? Will it be fun?

Some enthusiasts will surely always exist, of course. There are still people that ride horses for pleasure and for a living. But the industry won’t need them like it does now. Those who dislike or are indifferent to programming will have taken over.

Of course, this problem is larger than just programming, with all kinds of professions, hobbies and human activities in general suffering in much the same way.

This is just one possible future. I hope it doesn’t come to pass, because I, for one, enjoy programming.

[1] Reviewing code is much less enjoyable than writing code, and you don’t learn nearly as much. Letting AI agents loose and reviewing their changes is supremely boring.

Fringe cults have been talking about artificial intelligence in religious terms for decades. Over the last couple of years, it seems like the discourse framework, if not exactly the message itself, have become more or less mainstream (at least in the tech space).

No one expressed this sentiment more candidly than Masayoshi Son (Softbank founder and CEO), which a few months ago cautioned us of “blasphemy against AI”. I’m not one to question people’s piousness; however, its important to note that Softbank has a $40B stake in OpenAI, so maybe the man had some worldly interest in the matter.

A couple of days ago, DHH (Ruby on Rails creator) expressed a similar thought in slightly less theological tones in his Rails World 2026 keynote: no one knows what exactly is going to happen, but you need to be optimistic about AI (pessimists are “fucking losers”).

Obie Fernandez of Hashrocket wrote a piece in a similar vein: like DHH, he’s “proud of not reading code anymore” and cautions AI-naysayers that “their career is dead”, and “having a temper tantrum” is useless. They are powerless against the “tidal wave” of AI.

See the pattern there? Besides the lack of empathy and basic table manners which we’ve come to expect from tech people, the tone is always of cosmic inevitability.

In the tech space, it seems increasingly hard to find unapologetic criticisms of AI (yet alone simply disliking AI). Pointing out any of the many issues with AI, from the absurd capex to the catastrophic power usage increase and its effects on global warming, is frowned upon.

If some form of criticism is unavoidable, it needs to be prefaced with a Credo to assuage the reader that you are not a heretic: “Don’t get me wrong, I use Claude every day; but, …”. Only then is the criticism tolerable (barely). Negative takes are not allowed - unless you want to be labeled a “fucking loser”.

The dogmas are being formed, and questioning dogmas is heresy.