davydm

Personally, I stay away from ai-gen code because I don't want to debug someone else's code when I could just write my own. AI-gen solutions, from my experience, only fall into a handful of categories:

1. boiler-plating

imo, this should be done by bespoke tools, which are crafted by someone who understands the domain - and that's how it's been done for ages. If you can get accurate boiler-plating from ai-gen and there's no tooling to help - go for it, but you'll miss out on understanding what that boiler-plating does if you don't already know

2. trivial solutions

IE, solutions that are obvious, like "write me a function which determines if a number is even". Sure, this prompt came from someone who doesn't understand their own language and tooling, so it's already coming from a low bar. But the solution here is an obvious one, doesn't require a generated function, is literally built into the language

3. completely wrong solutions

The amount of time I've wasted reading ai synopses when googling a technical issue... Egad. Because they've all been a waste. I spent 1/2 an hour with chatgpt trying to resolve an issue with webpack, and despite me telling it several times that it was in webpack 4, it kept trying to offer incompatible solutions that work in webpack 5. This is just one example, but the number of times I've seen ai outputs and just gone "wtf, no", is not insignificant. These may sometimes be not as harmful as the next case, because sometimes even people with low skill in the area can spot the problems, or they simply don't compile, ie break before they have a chance to manifest havoc.

4. subtly-wrong solutions

This was my experience when being presented to by people saying "look what this model does". The problem is that others in the room didn't spot the bugs until I pointed them out, so I'm not sure if that's due to overconfidence in the output or they just have a skill issue. Subtly-wrong solutions without someone to correct them are the worst because they will appear to work most of the time and present weird bugs sporadically. One output I saw was around doing something based on the day of the week, and the output just skipped sundays, for no good reason (was counter to the requirements)

None of these outcomes are useful to me, and they eclipse, by far (in my experience), the maybe 1% of useful outputs. My conclusion is that it's more time-efficient to do the work myself, because I understand the long tail of debugging and extension and all the things the ai subtly cocks up.

Design is not just what it looks like and feels like. Design is how it works.

Perhaps, in that complex individual, there was a respect for Woz and his kin - the people who make stuff for other people to use, and delight in said use.

I can't wait for this shit to explode. I look forward to the ai crash. Some sanity would be nice.

It's your "average joe" users who are turning to "ai companions" when they feel depressed and want to kill themselves, because they are under the impression that it's an understanding, sentient entity. If people all thought of LLMs as "bags of words" (ie, understand that they're just token predictors with ZERO understanding of anything), perhaps that would change.

But we can't rely on the grifters to be honest - they'd lose all that yummy yummy cash! - so I guess we need regulation, just like we do for driving cars or taking medications, or any other item from a myriad I could choose from.

without openness, there's no way I'll trust some arbitrary iso on GitHub - 'tis a good way to have a bad time

I was mostly interested in how it's built, what the base is (debian/ubuntu? arch? gentoo? rpm-based?), but I won't be firing up a vm to "figure it out" - too much on the go.

But kudos on making something you find useful - that's how all good programmers start.

producing more code is not the same as being more productive

producing more working features that your users actually want and squashing bugs so that they have confidence in the product is productivity

good senior engineers should appreciate that an entry/junior (or even mid) dev may not "produce" as much - and should be helping you to upskill

producing large amounts of trash code that someone has to review and ultimately fix is not productivity - many "ai-first" companies are starting to learn that

so I'd suggest: focus on your craft, on the basic principles, but also the niches and the interesting bits. Find a part of it that you enjoy, that you're good at - preferably even outstandingly good at - and keep going.

apparently, this "you're a bot" fad is catching on - can't even access the site without archive.today :/ good job ai grifters.

no thanks, i'll keep crafting code by hand, test-first.

It's overconfidence from sycophants who don't understand that the very heros they worship either don't grok the limitations of llms, or outright lie about them. It comes from people seeing what's not there and extrapolating.

imo, agis will come, but not from llm roots. They may utilise llms but they will be a different thing altogether - one that is able to learn and apply logic and actual understanding to problems, not just try to spit out the next word that would commonly be there based on many many data lifts.

well done, asshat

I'm less fond of the bold statement: """ The jobs most affected by AI share some commonalities, said Zarifhonarvar. They are especially vulnerable because they involve tasks such as data processing, writing, analyzing information, and providing advice, all of which AI can now do better and faster than humans """

Faster? sure. Better? PFFFT HAHAHHAHA are people not paying attention to what these bags of words are spitting out? Every time I look at an "ai summary", I either find nuggets of pure bullshit in there, or, if it's actually accurate, it's "well duh", like telling me that the "WiggleFrobnozzle function wiggles a frobnozzle". Thanks machine. Really helping.

this is the end-game for ai - when every site fights to keep you out, they're all equally useless

there are a lot of factors involved, and AI is likely facilitating some of the bad behavior, but the problem seems to always come back to greed and corporate profit /2c

Back in the late 90s, I struggled to get my first job that actually used my studied skills. Yes, I had a job - delivering pizza - but I was desperately trying to get a job _coding_, and in-between virtue-signallers only hiring people of a specific race and the vast majority of companies expecting years of experience, it took quite a while and I really only found a position because the place I studied at gets actively involved in the process.

Yes, the trend of trying to replace entry-level jobs with AI doesn't help, but this isn't a new problem. Few businesses have the foresight to train up employees, and a lot are understandably disillusioned by entry-level churn, where you train people, and the moment they have skills, they skip off to another company, usually for more money - when losing that person costs the company more than if they'd just pay them a market-related salary.

AI isn't helping, but it's far from a new problem - it's just exacerbated by the corporate greed chasing ever more profit, where the easiest way to see a short-term financial gain is to cut salaries.

I'd say it's more akin to learning vim, where there's a fairly steep learning curve at the outset, with huge rewards down the line.

Yes, initial install can take a while. Yes, system updates take longer when built than from binaries - but modern hardware makes that a fairly trivial difference. Most people who have a problem with Gentoo either only used it decades ago when build time were a lot longer, or because they heard it takes long, which is way more common than the former.

Worktrees are unnecessary complexity when you could just clone twice. You can clone from a local repo too, then change upstream, if bandwidth is an issue.

Worktrees are bullshit.

i still blame the ai hype-bro grifters who are still taking major losses whilst they try to bait people into paying big dollars for over-excited auto-corrects.

In their greedy quest for ever-more money, these charlatans are the real grime in the system.

textarea { color: red; }

textarea.disabled { color: gray; }

Aiming for full semantic html and applying styles at the tag level _is_ good, but this seems like a not-very-useful refactor which has only made the code more rigid /2c