Many small oh shit moments, mostly of the variety of: "Oh shit, why am I still paying for this app subscription when I can vibecode it myself and just pay less than $1 per month in API costs, if even that?"
Overdramatic: when I saw friends and acquaintances doing this I couldn't help but feeling a slight sense of loss--that we (I) have lost the person.
At that point, is the person still even a person? He's nothing more but a meat RPA, copy pasting responses.
The reason I value a person is the uniqueness of the person's brain's weights and biases. When I lose access to that and I get ChatGPT/Claude/Gemini weights and biases, isn't the person... essentially dead to me and the world?
It's a very unsettling thing to think about. What makes a person a person isn't the fact that the person's breathing air, eating food, copulating, defecating, but it's the person's wetware's weights and biases. Because without those, what is even this meat construct I'm talking to via WhatsApp?
While I endorse the message of TFA (though do find the framing a bit on the overly blunt side), I believe it's unfair to reduce to "losing the person". The person is still willing to engage with you and still had to use their human words to prompt the AI. The latent space they exposed within the model is still uniquely the result of their words and effort.
We're just missing the establishment of a decorum of, "even if you do feel like you need to prompt the AI before responding, and even if you like the response, you still need to paraphrase and synthesize to avoid coming off rude and inhuman."
> The person is still willing to engage with you and still had to use their human words to prompt the AI.
But they aren't willing enough to participate to actually engage in a discussion. They're just making people interact with a bot with a layer of indirection. Apparently, they don't know (or care) that if someone wants the machine's answer rather than the person's, they'd just ask the machine directly.
> At that point, is the person still even a person? He's nothing more but a meat RPA, copy pasting responses.
You can say nearly the same of someone obsessed with social media and brain-rot. If you don't actively resist, soon your world view becomes the algorithm that you are being fed.
Suppose they simply directly quoted Wikipedia or some other expert authority, with attribution, rather than AI. Would you say the same thing? The person isn't giving you the uniqueness of their brain's weights and biases when they do that, either. That doesn't make the response any less helpful or appropriate in the situation.
People quote other sources sometimes. That's entirely OK. In fact, sometimes it's completely appropriate. We have to get used to the idea that sometimes, that source will be AI, and pretty soon (if not already) it will be just as authoritative and correct as Wikipedia or any other expert the person might quote.
If you don't like it, instead of responding by sending them a link to an aggressive, insulting, disrespectful and frankly low emotional IQ site like this, you can just say, "OK, thanks, that's great, but what's your opinion? I'm genuinely interested in hearing what you think." Unfortunately, if you send a link to this site, you are more than likely to lose the person entirely from the conversation anyway.
> People quote other sources sometimes. That's entirely OK.
It's OK in the context of doing that to support a larger discussion they're participating in. It's not OK if the quote is the entirety of the response.
> you are more than likely to lose the person entirely from the conversation anyway.
Perhaps, but they weren't really participating in the discussion anyway.
Internal assessment of self-worth is one thing. But one thing that I noticed while I was between jobs, was that the rest of the world was also built under the assumption that you "had a job".
Sign up for financial anything, they always ask you, which company you're with? What's your title? What's the range of your income?
I don't know if this is the case in the US, but in my country, I couldn't even open a brokerage account because the automated form required an office job. Entering freelancing or anything of some sort will get auto-rejected.
So it is in your face, all the time. And actually at that time I was fortunate enough not to have to worry about bills etc.
I'm about a decade behind you, but I also started my programming career during the "good" COM/DCOM/MFC/ATL/ActiveX/CORBA days. Java just came out. I slept little during that time because truly, there was nothing like programming. It was the thing that pulled me awake in the morning, and pulled me from falling asleep at night. I was so spellbound, calling it Csikszentmihalyi's flow felt like it didn't do it justice.
Fast forward 30 years later, I thought those days were gone forever. I'd accepted that I'd never experienced that kind of obsession again. Maybe because I got older. Maybe those feelings were something exclusively for the young. Maybe because my energy wasn't what it used to be. Yada yada, 1000s of reasons.
I was so shocked when I found out that I could experience that feeling again with Claude Code and Codex. I guess it was like experiencing your first love all over again? I slept late, I woke up early, I couldn't wait to go back to my Codex and Claude. It was to the point I created an orchestrator agent so I could continue chatting with my containerized agents via Telegram.
"What a time to be alive" <-- a trite, meaningless saying, that was infused by real meaning, by some basic maths that run really, really, really fast, on really, really expensive hardware. How about that!
I'm significantly younger but also programmer for two decades since my early teens and am experiencing something similar. CC is so freeing in that it makes those "nice but no time" ideas into reality by doing it next to your main project, almost feels like a drug.
It suddenly turns that dead time while you're waiting for CI, review or response into time where you can work on the fun or satisfying side projects by firing up a few prompts, check an iteration or 2, and then pause again until the next time or while the agent is doing its thing
Slang? The IDE looked like Turbo C++ of old (blue, text based interface). Shortcuts are weird, so you need to remap keys to get sane defaults.
Probably the most unique feature is that the language supports spaces in identifiers. So you'd have variables like "Option Portfolio Risk" or functions like "Calculate Estimated PnL". Visually obviously different from Python, but it gave me Pythonic vibes.
It's also nice that it supports preconditions, so you can specify the valid range of arguments etc. It has some kind of OOP support but tbh it felt bolted on (understandably).
But the most value adding, IMHO, is the DevEx and deep integration with SecDb. Say what you want about the DOS-like IDE and the old (20+ years old for sure, maybe 30+) language, but you can deploy your code SO easily into production, with guardrails in place.
Out of curiosity, I implemented a toy language (thanks to Robert Nystrom's Crafting Interpreters) that supports spaces in identifiers (https://github.com/rayfdj/gaul-lang) as well. Makes for an interesting weekend coding project, and it helps me understand more the tradeoffs that Slang designers must have gone through.
fn Calculate Portfolio Risk(Initial Investment, Risk Factor) {
let My Very Special Adjustment = 0.95
Initial Investment * Risk Factor * My Very Special Adjustment
}
that is so cool; this is actually something i've been looking for a long time
and jam karet looks interesting; `if input ~= "yes"` made me smile.
i also liked the keyword replacement for multiple languages as well, that could be super usefull for children learning programming i'd think!
I love how uv allows me to not think of all the options anymore.
virtualenv, venv, pyenv, pipenv... I think at one point the recommended option changed because it was integrated into Python, but I can't even remember which is which anymore.
Such a pleasure to finally have just one, for maybe... ~99% of my needs.
The fun / creative part for me is not googling “how to slurp the contents of a file into a string” or “the exact syntax for marking some functions as unit tests” or “the correct order of symbols to specify generic type param”
It’s not “the correct html / css syntax for this basic gui I want to make”
It’s not “how to achieve the thing I’ve done 10 thousand times in other languages/frameworks, but for this language/framework”
It’s figuring the core logic out, building the thing while skipping the boring stuff, playing with abstractions that scratch my itch.
From this pov, AI is the best thing that has happened to my weekend coding. I code recreationally way more than before. Before AI, I would try a new language or framework, and I’d give up halfway because re-figuring out basic stuff for the umpteenth time is boring, it’s not fun at all. Now AI lets me skip those boring parts.
Learning Elixir and fixing a bug in an open source project went from "risk of a long slog over the course of a month with no reward" to "pepper an LLM with questions (debugging errors, understanding syntax, translating code snippets to English descriptions of behavior), write 20 lines of code by hand, write a few test cases, and submit the PR fix".
Strong agree. I've been making more progress on my passion project in the last few weeks than I have in a year, because it helped me break out of analysis paralysis.
I'm really, really, loving the agentic flow, where it digs itself out of syntax errors and the like.
Current tools: Visual Studio Code/GitHub ChatGPT5(Preview)
An example I’ve done with a small language model is fine tuning a small model for evaluating semi structured user input for whether it’s likely sanctions evasion. The heuristic code to do this due to the fact is got a natural language component is very complex and has a relatively poor precision and recall. A fine tuned model pretty reliably classified the inputs and was pretty resilient to adversarial attacks while heuristics - partially due to their full determinism - were very brittle over time.
These sorts of “heuristics are surprising incapable while the semantic flexibility of language models are powerful” are surprisingly large. Even flexible validation mechanisms that take human entered semi fixed form inputs and reformat them to the expected input in a really reliable way and is much less frustrating to end users. Essentially any situation where abductive logic or natural language comes into the picture a small model does really well.
Both of those things - abductive logic and natural language - were largely unavailable as tools until recently. This pretty nearly rounds out the complete toolkit for making really robust and powerful (and usable) systems. You sacrifice a perceived determinism by admitting abductive logic and non determinism, but in my experience this warm blanket of the mathematically inclined wasn’t particularly robust in reality and systems often deterministically failed in complex and difficult if not impossible ways to avoid that a little bit of abductive reasoning could make it remarkably simpler and more robust.
I've been taking these rides 5-6 days a week, everywhere, and also in other countries outside the US. What I've come to realize is this: what matters to me the most is the consistency of the lowest bar of the experience.
I get that sometimes with human drivers, when I'm lucky, I get someone who goes above and beyond, someone who's fantastic to talk to along the way, and so on.
But if I can trade all that with a guarantee that there's a consistent, predictable floor to my worst experience, I'll take it in a heartbeat.
At the end of it, I take a ride to get from point A to point B. I'd rather have a machine does it for me very efficiently, without all the messy human element, with the ups and the downs, because it's the downs that ruin my day.
100%. I've discovered the floor: Small cars that probably aren't safe, trashed interior, and drivers who smell of literally every vice while talking on the phone AND playing whatever music all together. "Premium" is simply not that experience.
reply