I am sick of talking about AI — literally feeling sick and anxious when it comes up — but there’s something I need to unpack and find words for that’s more than boredom or fear. So here’s more words on the topic that I’m sick of.
While a friend first showed me GPT 2.0 back in 2019, it was only February last year that it became a focus for me and my work. I read Harper Reed’s “My LLM Codegen Workflow ATM”, experimented with the early coding agent Aider, and it promptly became my team’s job to figure out how to help my company adapt to this future way of creating software.
So it’s only been 19 months, but I’m exhausted.
I know the old Upton Sinclair quote applies: “It is difficult to get a man to understand something, when his salary depends upon his not understanding it.” The problem is, I’m not even sure which direction it applies in: am I avoiding AI because it’ll disrupt the career that’s been both enjoyable and highly paid? Or is my well-paid job helping my company adopt AI effectively the only reason I haven’t completely walked away from this technology in frustration?
There’s things to love – like being able to prototype so quickly!
It’s also thoroughly disappointing. It’s not reliable at a quality level, and it’s fundamentally unsafe at an agency level: you’re never sure if it’s following your instructions or someone else’s, so the amount of speedup you can achieve follows the amount of risk you’re comfortable living with.
And there’s the things to hate. The work-slop, colleagues sending a document they didn’t even bother to read themselves, and boasting in the efficiency of that rudeness. Experts being asked to look the other way because someone less qualified has vibe-coded a thing and we don’t want to slow down with difficult questions. The article “Appearing Productive in the Workplace” resonated with me deeply.
But I think more than love or hate or disappointment, I’m feeling grief about what we lost.
I’m grieving the loss of programming as a paid task. There was a satisfaction in writing code, getting it to compile and the tests to pass, and seeing it work. It was rewarding, enjoyable, and a skill that paid well. My life doesn’t allow for many hobbies, so doing something enjoyable for work was huge for me. I already miss it. Realistically: I was already leading teams and not individually writing code as much, and like many engineering managers before me, I was already on the path of programming no longer being the main job. But I’m sad that it’s not really a role I can go back to.
I’m upset that the care for our craft is disappearing. Anyone can create software now, and even if I can do it better than someone without training … if they can do a good enough job for almost nothing, does our expertise matter anymore? We used to have a “Front End Practice” meeting where we celebrated the craft of building for the web and learning for each other. But attendance fizzled out (myself included), and it’s now been cancelled in favour of AI peer learning. I miss what we’ve lost, even though I have myself to blame.
I miss talking about niche programming languages and frameworks: better abstractions to help humans write better code. Coming up with the right abstractions was rewarding work: it encouraged deep understanding and deep thought, and when you got it right, the pay-off was amazing. The incentives to do this now are all wrong. The new language won’t be in the training data, the agent would be more likely to succeed if you stick to Javascript or Python. I fear we’ll hit a local maxima, and I’m sad that those fields of discovery that I loved exploring look likely to fade out.
I’m sad that budget for team building is evaporating while budget for AI token usage appears to have no limits.
I’m sad and angry that we’re not good at talking about the environmental impacts. We have thought leaders saying the future looks like AI spend at US$1.4m a year for a single engineer which my napkin maths says is the equivalent of 157 households electricity use in my home city. I’m upset that I’m resorting to napkin maths and a lot of assumptions for this, and that people actually in the industry aren’t talking about it.
I’m grieving that we used to dream of making the world better through software. It was a bit over the top — and rightfully parodied — but I was one of the people that did choose what I worked on based on if I thought it would help make the world better in some small way. All that sense of “make the world better” has been replaced by “ship faster”, but this is meaningless if we’ve lost touch with building something meaningful.
It’s upsetting that the idealism of the early internet: decentralised and non-corporate, community driven and open-source, aiming to improve the world, has largely been lost. Now we’re on a path to forgetting how to build it ourselves without a subscription to these massive AI corporations.
So I am sick of talking about AI.
And I am angry at how it’s reshaping our world.
But I’m also grieving what has been lost.
I liked this industry.
I liked the internet we had aspired to.
I’m grieving.