HI version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
56% Positive
Analyzed from 6370 words in the discussion.
Trending Topics
#don#water#more#code#lot#using#years#system#world#things
Discussion (129 Comments)Read Original on HackerNews
Me, I pointed Claude Opus 5 at a C codebase I've been working on for years and it immediately found five serious bugs and told me how to fix them, as well as how to use Linux system calls I wasn't familiar with to solve some other problems. And I gave it a new language for PEGs I'd written up a couple of years ago, and it wrote me a working implementation in OCaml that afternoon, fixing several bugs in my example grammars along the way and resolving a conceptual problem I'd had a roadblock on. I asked Fable 5.1 to implement a minimal proof assistant, and it wrote a clone of J-Bob in Python, which I'm studying now. Although there is a certain firehose quality to this stuff, I sure don't intend to use AI to actively de-skill my brain.
That doesn't guarantee that AI will be a beneficial innovation overall, of course! As with most innovations, it probably depends on the balance between people being able to use the innovation to gain control over their own lives, and people being able to use it to gain control over others' lives. (Barring a FOOM scenario, of course, where our prediction ability is nonexistent.)
Basically, I don’t know how well thought out it is, but the prose argues that the point isn’t for it to be an objective take, and that if it were well thought out (ie highly iterated on and more objective or factual than subjective) some personal meaning that was intended to be expressed is lost along the way.
In other words It may have not appealed to you but that’s fine - you might just be outside the target demographic. In any case I don’t think it necessarily should be more well thought out. I prefer that these pieces exist or have their place. I definitely prefer their existence to a world where every personal Blog post has to arise from fiduciary or technical concerns, instead of from the soul. This is a soulful blog. I like it.
p.s. and at the same time, it's such a fantastic learning tool. It distills the intuition of a whole world and is able to transmit it on demand, like in your examples. It's an interesting dichotomy.
The only problem I have at the moment is it's very prone to hallucinations, even now, yes frontier models astra fable with all the bells and whistles. This makes it hard to confidently use while learning because I have to be on the lookout for lies while I'm learning, which is precisely the moment I am least able to distinguish them. So instead theres just a constant low lying dread.
Nonetheless, I am able to get some value out of them. Just not all, everything requires manual effort to duplicate and check which you should probably be doing anyway as part of learning
Structurally, though, I'm sort of pessimistic. Throughout my career as a programmer, I've earned much more money than electronic engineers whose skills are mostly very similar, and probably much of this happy situation is due to the fact that my productivity depends only on me having access to a computer with GCC and Perl or Python on it, and the internet. If I were living in rural Malawi, I probably couldn't practice my trade, because the US$200 it takes to buy a working laptop is a year's wages there, and people lack the social permission to innovate. (William Kamkwamba's astounding achievements under those circumstances deserve more recognition. Read his autobiography!) But, if I work in any developed economy, that's somewhere between 1 hour and 20 hours of my earnings, so I'm not dependent on investors to create jobs for me.
By contrast, electrical engineers mostly can't ship their designs unless an investor is willing to put up the capital to produce a batch of PCBs or chips — probably three batches or so before you get a fully working product. If you're working on a cutting-edge chip, that's several million dollars, so you're probably not going to save that up while working as an employee. This puts electrical engineers in a much worse bargaining position relative to investors. To a smaller extent, the same thing applies to machinists, auto mechanics, and many other skilled workers: they only get jobs if investors invest to create jobs for them.
The capital-intensive nature of current AI seems like it might push the negotiation equilibrium for programmers two or three orders of magnitude in favor of investors.
But it might not work out that way, and, at least at the moment, I'm finally bringing about numerous projects I'd been procrastinating on for years.
(... Palm Springs has around 100 golf courses)
Likewise, you may not feel like AI is de-skilling your brain. But it probably is. We don‘t have the research (yet) that AI is addictive and causes brain atrophy, but all the evidence is certainly pointing that way, the same way the evidence were pointing that way about tobacco in the 1950s. My only hope is that it doesn’t take us another 2 decades to prove the obvious.
It is, but why is that a problem? Moving from physical labor on the farm, factory or construction site to sedentary office work and lifestyles would also atrophy our muscles so now we pursue physical activities we actually enjoy like gym and sports instead of hanging on to those dangerous and strenuous physical labor jobs just to stay in shape.
Similarly, you will use the AI to churn through the repetitive mundane tasks the AI can do, and then work your brain on the activities you actually enjoy.
Now if you choose to not exercise your brain and dumb yourself down the same way sedentary people don't exercise their bodies and get obese, then I guess that's on you, but we can't force the world to technologically regress just because some people are lazy and don't exercise, the same way we haven't forced construction workers to use pickaxes instead of jackhammers to tear up a road just to make sure they stay in shape.
I think that was the same day I prototyped an approach to using Service Workers from localhost as a platform for persistent mobile apps that don't show up in the app list but also don't rely on the TLS CA system for their security. Which involved learning a lot of stuff about the Service Worker lifecycle and localStorage and IndexedDB that I didn't know before.
It's just a lot.
When I see people worry about LLMs atrophying their mental abilities I assume they haven't thought very hard about ways to lean into using them to expand their mental capabilities.
These 'serious' bugs aren't important if you've been working on this code base for years and only now they pop up. I suspect - although I'm guessing - that probably a hobby project.
Frankly speaking, although this comment is very dismissive about the opinion blog article - assertively stating that this isn't a well-thought-out post - it seems to me that the personal anekdote is irrelevant and that the blog post in question still fully stands.
Sure they can mostly write better code than a toddler but this constant nudging and reminding and reiterating and stopping them from using the token budget of the whole company for a one off script. It gets tiring and I feel like I am losing brain power while doing it. Maybe it's faster but explosive diarrhea is also a faster way to produce shit.
i'm in the midst of a very large global roll out of a new system including about 24 separate erp integrations. My team's system is dependent on another system coming online first then we go then the integrations come online then a massive data migration happens and ends with all the associated testing. So like a huge number of dependencies and I'm second in line. Today we get word that the first system is having issues with a vendor and will not come online as scheduled (our dry run deployment is tomorrow and spans all next week as each dependency gets their turn. Week after next is PROD.
We're not heavily integrated into that first system but we are in spots and so there's a reason why they go first. Writing around that system is feasible but we're looking around 5k lines which isn't a lot but there's still testing and sign off. My offshore team is in bed, meanwhile they, and onshore, all have other shit to do anyway so pulling them just puts everything else behind.
I tasked an agents to make the changes as soon as i woke up to the news, they got it mostly right. It took two rounds of fixes with some offshore testers who stayed up late but the PR got in and merged. We just did (literally like 5min ago) a validation run against our target system and we're green for tomorrow. I told the testers to goto bed and then i pulled up HN and opened a coke.
There's no way in hell that would have happened without the coding agents we had on tap. So, yeah, i find a lot of value in the rate at which AI can work on code even if it's not perfect every time.
Proof of this is the amount of unnecessary tasks that Claude does just as an excuse for not doing a good job doing the tasks that we ask it to do
I'm still a lot faster for certain things with AI but its weird. Instead of just being annoyed about some weird bug, i'm annoyed about an LLM which can do everything and then falls over itself on weird issues.
It feels weird true but it still also relevant faster?
It's called a fancy word "steering". Most of engineering is now steering or providing "taste" to AIs so they don't produce slop
I also found that slowing down your development (just by a bit!) on purpose helps. Do not rush into implementing your new idea and give yourself at least a couple of days to think it through.
Also we’re just so damn stupid to learn from history, or even just people 10 years older than us, that we repeat the cycle.
And there I guess is where saying a new sucker is born every minute is from.
There's the lingering societal PTSD to consider, but I think most of the challenge in the aftermath are latent issues that preexisted covid and it just brought them to the surface. I think our world a couple decades from now will be stronger than it would have been in a universe where covid never happened, and actually has a better chance at dealing with the next covid-level threat. Perhaps that is this, in a way.
Covid proved that a healthy looking society is not always so healthy, and I think it works in reverse too. The outcome of the upcoming election will tell us something about how far we have to go.
The places that needed restrictions such as hospitals & nursing homes still got covid by & large. We didn’t necessarily need to close down the whole show.
I’ve had my jabs, also had covid multiple times. I’m not a disbeliever, I just think the collective reaction was completely OTT.
As a result, they had relatively short periods of lockdown compared to other countries, such as the UK, which had comprehensive vaccination but dithered about going into lockdown.
I first started noticing this in the 90’s .. comp-sci grads were entering the industry with no clue whatsoever how to do software. (Of course its gotten worse since then, but I started noticing it in the 90’s..)
Everything old was thrown away as being archaic and obsolete - only the New Cool Shit™ mattered .. except nobody had ever done anything, properly, with the New Cool Shit™ - everything was still running on the old.
(Unix was dying, everyone was learning only Microsoft tech because “that was the future” ... until it wasn’t. Now, a majority of us are running 21st century Unix workstations given us .. by Apple, of all companies, which didn’t die so easily, after all..)
People entered the market not because they were truly interested in improving the world through computerization, but because they wanted the lifestyle afforded to them by a career in “Information Technology”, a nascent term back then which simply meant “career path for the budding materialists”. Trouble is, you can’t be materialistic with computers, they are almost entirely a spiritual device, which value is determined strictly and only on the decision of value made by humans. It is humans which decide the true value of all that physical hardware after all.
I think this cyclical amnesia is driven by materialist consumerism. People who take a less material approach to technology, tend to build stuff that lasts a long time.
The reason I think it happens in cycles of 4 years, is that is the average time it takes someone to enter the academic churn, and get spit out on the other end, ready to join the work force and do big things.
Or .. not.
They know what they're doing, the people who fund them know it and the people with half a working brain cell sooner or later hear someone like me and go 'oh, that explains it'.
It's much simpler than some mythical cycles - manufactured 'mass stupidity' is a conscious, deliberate, ongoing choice.
It used to be religious organizations, then it was newspapers, now its social media platforms. The playbook has been the same for thousands of years - nothing has changed.
To save everyone some time - the playbook is you're cattle, I'm shepherd. I'm master, you're slave. It's not anymore complicated.
The bit 'educated' people miss is that most people would love nothing more than to be the master (freedom they call it? American doublespeak). They dislike being slaves, not the concept of slavery.
Anyhoo.
Good try.
This rings true.
I don't know how this will play out, but I haven't seen anything like this before.
The first group is generally disillusioned, for two reasons. First, it's an attack on their hobby. Second, they sort of can't let go: they want to own the product, they want to understand if the code is good or bad, and they spend a lot of time cleaning up after LLMs. Which, frankly, is not fun.
The second group is over the moon about it because they feel it gives them the elite powers previously only reserved for greybeards.
I don't think this is unique to coding. It's the same thing with writing. If you like writing, you probably hate LLMs. If you see writing as a boring chore that needs to be done, you are delighted to have a button you can press. Haha, take that all the influential bloggers, tech writers, etc. I can now do your job at 10x the speed.
Just a bunch of controversial ideas: structured programming, null, intellisense, microservices, cloud first, serverless, OOP, "Clean Code", exceptions, SaaS.
When I tried using agents for something as a test it immediately started doing things instead of being part of a conversation. I want to be actively involved in the process, not sit back and let AI agents write/generate stuff that I'd have to/end up rewriting/modifying anyway when it goes against the design I have in mind.
The other thing I don't like about agents is their ability to run any command [1]. That seems like a nightmare w.r.t. the potential for leaking secrets (signing/access keys, etc.) or doing damage (deleting files, database tables, etc.).
[1] You can set the option to review every command it runs, but you're then just hand-holding the agent.
The approach currently being foisted upon me is to offload almost everything to LLMs (have it write a complex tech spec -> one shot into tickets -> one shot implementation of said tickets), with humans reviewing throughout, and repeating this iteration until it 'looks good'...throwing sh*t at the wall until something sticks, basically. However, we're being expected to review at a far faster pace than one reasonably can while maintaining an understanding of the system.
Meaning, with this paradigm, humans will largely be rubber stampers and will atrophy their abilities as a result, which will seemingly result in compounding adverse effects.
TLDR in my experience the AI can help with some 'busy work' eg translating a well defined solution in my head into code. The time it takes to understand an implementation and ensure correctness will not change. LLMs cannot reliably do this. This traditionally has occurred during implementation by the implementer, and during code review by code reviewers. I don't think simply shifting the implementer's comprehension 'duties' to code reviewers provides any benefit.
It was a huge shift in the world of system administration, gave birth to devops, and, along with the rise of containers & cloud computing, it radically altered how companies ran their infrastructure
Some people leaned into it, others didn't and a huge rift opened up between them
I used to say "automate all the things" - I still do, but I used to too
AI tools are just an acceleration of that mindset and in my world of cloud/devops/sre I'm heavily leaning into AI-driven development and daily workflow
I still count myself as "dejected" overall. I'm dejected by the economic shape of AI which seems destined to keep concentrating outsized rewards to the top few who own it or have enough capital to shackle it to their will. I'm fearful that intelligence, once it becomes a commodity you can rent any time you may need it, will be socially devalued in our already anti-intellectual society. I worry that in a couple decades we may resemble the world of "The Machine Stops" (1909), where nobody really knows how things work.
Of course we already live in such a complex world, no one person has complete knowledge of the "stack" they rely on. I can write software in C or assembler but I can't design a CPU and don't really understand how one is manufactured. (I may someday study that field, but I haven't yet!) But I take solace in the fact that somebody out there, has put the time in to study each of those things and is an expert in them. At every layer and in every niche of our engineering world there are masters of their craft, and there are students learning the fundamentals to keep that knowledge alive and growing. What if in 30 or 40 years that is not the case, and there are whole corners of knowledge the modern world depends on where everybody is reduced to "I dunno how it works, but Claude said..."?
Oh my god this nails my experience to a T.
Now a new powerful tool shows up and I think it is time for the Jesus Christ camp to throw in the towel.
AI , trained on that garbage , is producing what now? Supposedly clumsy , buggy and so on.
The lack of self awareness is just hilarious.
Thats just not true.
I only wanted to say that finding security issues was def bottlenecked by humans.
But if you do think vulnerability research matters, and you're trying to argue that frontier models aren't a seismic change for that discipline, you have almost no company. Vulnerability researchers are overwhelmingly leaning on automation to find vulnerabilities and, just as importantly, generate the tooling required to test hypotheses.
You can feel about that however you want to feel about it. I mostly don't care, except: you can watch people like this being negatively polarized back into the bad old days of the mid-1990s, content-free CERT advisories, and vendor-controlled "responsible disclosure" by a use case that frontier models unimpeachably excel at.
And yet we continue to do so. I understand the feeling of loss some people may be facing. And there are definitely some really bad practices - like nakedly spewing claudspeak at your colleagues instead of communicating. Or raising a PR you don’t understand. There are asymmetries we haven’t learned to navigate. But we aren’t returning to a world where it doesn’t dominate our discipline so it’s best to find opportunities.
For me the answer is still absolutely resolutely “no”.
Why is using AI tools self-debasing or degrading?
So it should be utterly unsurprising that many people are going through a bit of an existential crisis around this. Because what you actually sat down and enjoyed doing, what motivated you, may be entirely gone now.
And on top of that, there’s an extreme amount of pressure to do more in less time, which is by definition stressful. And do much more context switching.
LLMs are actually producing a lot of pretty good stuff, can debug things that would have taken an entire team a week, and the people running AI companies are also generally like the most boring supervillains imaginable. All true.
It is also completely screwing up the global economy, and most people barely understand what it is and what's happening.
It's also not going to collapse anytime soon as a technology despite what the economy does, and it's going to get a lot worse before it gets better. Everyone who's extreme on it on either end is generally wrong.
Though if any followers still exist, they probably wouldn't agree that 𒀭𒂗𒆠 was just a fad.
rotary telephones and pay phones went away but we still have phones.
What is screwing up 'the global economy'? LLMs?
Number of years of study for each would also be illuminating. Online content excluded.
It was annoying before when I had a very clear idea of what I wanted and yet it still took a whole week to implement.
Now, with AI coding, I can implement any decent sized feature, exactly how I want, in a few hours and it's all polished and ready, all edge cases handled.
Everything it's generating looks almost exactly like what I would have written, just written faster.
Personally, I feel avenged by AI because I had been saying for years that you don't really need static typing once you reach a certain level, getting types right isn't the most difficult problem and isn't a significant source of bugs. Everyone was gaslighting me about this... Yet now everyone can verify themselves because AI has reached this level I was speaking of.
Now anyone who coded with a frontier model on both TypeScript and JavaScript can see that it makes no difference at all. I can't remember ever seeing Claude mistake the type of a variable with either language... And TBH, I find Vanilla JavaScript from AI to be more succinct, not over-engineered like Typescript. Which is a point I'd been trying to make for years about incentives and actual patterns of usage (as opposed to theoretical benefits).
Same. I quit, aside from two contracts I'll serve until they run out, I'll be studying Electrical Engineering. Even if AI takes over that space... good luck constructing a robot with the dexterity required to pull cables.
To anyone in IT... go and study for some qualification in tasks not easily doable for a machine. The trades are going to be a goldmine especially with all the boomers retiring. Or buy some ranch in the midst of nowhere and farm artisanal chickens, cattle or goats.
(For legal reasons, the latter was a joke. Farming is extremely risky, requires a lot of upfront cash, a lot of what used to be generational knowledge, but it seems to attract a lot of IT people, especially for whatever reason the goat farming)
It's a great time to be alive. Everyone is infected with pessimism.
The main cause of water fear mongering is that one woman’s well was dirty during construction of a data center. First construction is not the center running, and having lived on a well they get dirty for all kinds of reasons.
https://www.eesi.org/articles/view/data-centers-and-water-co...
According to EESI (a bipartisan-founded non-profit), a large data center:
> can consume up to 5 million gallons per day, equivalent to the water use of a town populated by 10,000 to 50,000 people.
Also noting that AI datacenters can use more.
I don't think 'anyone who thinks X is Y' is a good basis for discussion, which is the real point.
> Approximately 80% of the water (typically freshwater) withdrawn by data centers evaporates, with the remaining water discharged to municipal wastewater facilities.
I don't disagree with your assertion; however you do have to realize that their grievances, while not completely informed, are not wrong.
In a world where everybody seems to be quoting grok with no further thought process you would think people would be more welcoming of someone who is "slightly wrong" on occasion. It's the strongest indicator that a flawed human being got their information not directly from a machine.
However I live in a much nicer area than Chicago and any Walmart, large mass of people, or DC would really destroy the PV as well as it would be an eyesore.
I don't really understand why so many just want to over industrialize so many beautiful places so that a corp. can sell the next gimmicky iteration of our "best model yet" which is just slightly slower and more expensive then the previous.
There are certainly problems that can be crated by dumping heat into a body of water, but that's not the argument being made.
I'm willing to bet that a datacenter creates a fraction of the pollution of a similarly sized industrial facility such as a chemical plant, refinery, smelter, or steel mill.
That source of power, since our grid is so outdated and struggles with the weight of these power-hogs, EVs as well as increased usage of AC during the summers create some dangerous conditions.
There is the risk of heavy metals getting into the water supply as the copper degrades but that will only be a problem down the line, that's prescient but not yet realized.
> I'm willing to bet that a datacenter creates a fraction of the pollution of a similarly sized industrial facility such as a chemical plant, refinery, smelter, or steel mill.
I agree; however I would not want a DC or a refinery, smelter or steel mill anywhere near my town. We have industrial districting for that reason, and because a lot of DC projects want to be put next to lakes and water sources, this causes issues for the residents.
https://sustainability.google/reports/google-2026-environmen...
https://www.aboutamazon.com/news/sustainability/amazon-data-...
Or about 0.6% of the water used to grow almonds
Fortunately, now we have computers, which are great at unit conversion, so we can cut right through these feeble attempts at obfuscation. In non-medieval units, 10.9 billion gallons per year is 41.3 million cubic meters per year or 1.3 cubic meters per second. The river a few kilometers from me has a discharge of 22000 cubic meters per second (of fresh water dumped into the sea), so Google is the equivalent using 0.006% of the Rio de la Plata. Lake Superior contains 12070 cubic kilometers of water, which is 1.207 × 10¹³ cubic meters, also of fresh water, so Google's current use would dry up Lake Superior in only 290,000 years.
Google is mostly not very close to Lake Superior or the Rio de la Plata; a more relevant river might be the Sacramento River, which drains into the San Francisco Bay where Google was born, and it's much smaller than the Rio de la Plata. Specifically, its average discharge is 797m³/s, so Google is using 0.16% of the discharge of the Sacramento River. (But some of that is at datacenters far from California.)
https://en.wikipedia.org/wiki/Water_in_California#Agricultur... tells us that agriculture in California uses 34.1 million "acre feet" of water per year (Americans will do anything to avoid using the metric system, especially when they're trying to be sneaky). In non-medieval units, that's 42.1 billion cubic meters, and per year, it's 1330 cubic meters per second, almost exactly 1000 times Google's use. Of that, 18% (240 cubic meters per second) is spent on alfalfa, or was in 02014, anyway.
One major difference is that you need fresh water, like the water of the Sacramento River, to irrigate crops, but you can cool data centers with salt water from the ocean. It's less convenient, because there are a variety of annoying things that happen with biofouling and corrosion, but that's what you'd do if hyperscalers scaled up by another factor of 100× so that their water usage became large enough to be environmentally significant.
But that doesn't happen until they have 100× the computing power they have today. So anybody who tells you AI datacenters are a major water user is lying.