alt Hacker NewsLLMs are eroding my software engineering career and I don't know what to do
Comments
This is good. We want less barrier to entry and more competition in software.
One other thing I find it is bound to happen is that this domain knowledge you speak of is just going to shift towards LLM domain knowledge.
Look at prompt engineering, and how quickly it became a hot thing. Does everyone know to steer their AI well? There's only so much a harness can do for you once you start attempting to one shot with a single sentence of 4 words.
As others said, "write a Rust compiler make no mistakes" can only work if you overfit a harness to that single prompt. Nobody is going to do that.
So the part you mentioned about the knowledge you accumulated around how to know that "trade-offs between implementations" and "idempotency to prevent double-charges" is just moving to the domain of the english language and tokenizers. One could argue here that this is far more interesting as it requires you to explore deeper into how we communicate and describe the world around us. Reminds me of physics and math.
I think there's an optimism lenses to it if you can grasp it as an opportunity rather than an inevitable doomsday apocalypse.
While LLM's are beyond junior level at this point, they're still just that. I don't really agree that the first two pillars have been affected.
I've shared a story before that between now and 2 years ago a developer who solely relied on AI has produced the same hot garbage instancing system within the same time period. For example back in my day in 2 years I went from writing a system that struggled with few hundred players to one that could handle thousands and far beyond that. The person using AI 2 years ago wrote a system that didn't work and wrote a system 3 months ago that doesn't work.
Everyone is saying how great AI is, but they're missing that the driver is just as important AI wouldn't be able to achieve any of this without capable (often seniors) using it and giving it guidance. It's really a difference between "it works" and "it works without flaws".
Of course AI can produce things that also "work without flaws" with solved problems and someone "recreating" something that already exists with AI is not that special, a junior developer could accomplish the same thing given the time.
But I do agree that AI becoming part of performance reviews and all that is producing more productive developers which is going to drive the cost way down. In a way AI is stealing from a developers salary and giving it to the AI companies which is pretty ironic considering how cold developers seem towards artists.
LLMs are good at general solutions but not specific solutions. As industries evolve and laws, regulations, and practices change LLMs will struggle because those things are not included in its training set yet. We'll always need humans to push companies in new directions in order to compete, unless we eradicate capitalism altogether and then we're all out of luck. No competition means no incentive to try and be better than the next guy which means no new products and services for humans to develop that AI hasn't seen already.
I don't know how else to say this but LLMs are just word calculators. They are becoming better for sure but at this point even Claude 4.8 is absolute shit at any complex task in a not so common field. I have been working on terrain contour matching algorithms for the last few days and, oh boy are the predictions about AI taking over the world wrong. Its the highest level of bullshit I have ever come across in my life. I ended up writing 100% of the actual algorithm myself. It's a productivity mess.
LLM is a powerful tool but it still doesn’t have the context that a person would have. A million tokens is a drop in the bucket compared to the overall context that the person guiding the LLM needs to keep it on track and being productive.
If you’re not a good engineer and you don’t have the domain knowledge, your token costs will be very high for whatever gets shipped, because you won’t be able to provide the context necessary to prompt machine efficiently.
Claude will still very often hallucinate bugs, explanations, domain requirements, that have no basis in reality. It will offer fixes and improvements that are pretty standard but not optimal. This is correctable if you catch it, but you need to review every line of code and comment, because in addition to being obviously wrong, it is often very subtle in the wrongness. For every bit of “slop” there is almost microslop, the places where it just kind of confidently guesses… and doesn’t tell you… but sometimes is correct anyway.
The “problem” is there’s less low hanging fruit. You have to know a lot to add value beyond being a middleman gating the slop. You have to really pay attention to the details to find some of the errors that it’s making.
Advertisement piece for the IPOs. We get this multiple times daily to pump the stocks and demoralize programmers.
Hear me out: what about just refusing to use them?
why would i ever want to use a tool that remove the part of my job that brings me joy? Fuck productivity, we were already doing good, when we were able to actually do our job, i.e.: not wasting hours in useless meetings, or doing customer care to idiots who could not be bothered to follow instructions, which i shouldn't be doing in the first place. let the LLM do that, or let the human assisted by the LLM do that. Not my job.
Code quality and architecture still matter, because they also make it easier for LLMs to reason about the system.
That said, Opus 4.8 and Codex 5.5 both can write code that is higher quality than your average engineer. They are not quite there yet in terms of code re-use, but I think that's a solvable problem.
I'm thinking about taking a plumbing course.
> when I step outside my area of deep knowledge I can no longer call BS on the agents
It's still funny that 4 years into this mania the models can hallucinate basic ground truths, humans are increasingly not reviewing the output, and misusing LLMs where simple automation would suffice.
My wife does project management and works with a lot of tech leads. They came to her with a project plan deck, and she started questioning some weird dates.
The LLM was able to pull artifacts out of their issuer tracker, but it just.. hallucinated some of the dates in the process of creating a project plan deck out of the underlying data. These guys didn't care to review and notice, and who knows what else it hallucinated content wise. They were happy to send this project plan multiple levels up the food chain with hallucinated unreviewed dates.
5 years ago they would have just written a script and had none of this mess.
The title should be: LLMs are changing my software engineering career and I know what I should do
> Of course, I'm still employable because someone has to review the code and steer the robot. But I'm just another off-the-shelf engineer now. I have no domain expertise that another Sr. engineer steering an LLM cannot match. All my finance and payment domain expertise, all the debugging intuition and distributed system knowledge earned through hours of sweat and tears, is now promptable.
Ownership and responsibility are the new currency for the engineering staff. Willingness to implement these tools and then own the consequences of their use is what leadership is looking for. They want their cake while they eat cake, and they will keep those around who enable something approaching that experience. Owning the side effects of LLM use is more challenging than our own natural output because of the radical volume increase and unfamiliarity with low level details. However, I argue it is still possible. It has always been significantly more expedient to poke holes in someone (something) else's work than it is to perform that same work. And, the executives know this. They leverage this capability too.
The relationship between the business and the development team has been tenuous at best. I've rarely seen a technology team that was properly subservient to the business that ultimately signed their paychecks. I every case I have personally experienced, it is was like a hostage situation where the business owners are in constant terror of the technology people screwing them over in some infinitely nuanced way they or their lawyers could never understand. Many business owners are looking at this technology as a way out of the hostage situation. They noticed a window that was left unlocked. They are going for it right now. Whether or not they will succeed in their escape is a separate matter. Whether or not them being held hostage was justified is also a separate matter. It really helps to keep these things in their own lanes.
“even though you're delivering code at a good pace, you're taking too long to deliver those Design Docs. Are you using AI? You should use more AI.”
This here is the crux of it I think… it’s often promoted that AI will give us the time to do the “real” engineering work of designing systems and really serving the user, but in practice all I’ve seen is further attempts at optimizing every last process with AI - just homogenizing every product and feature into slop.
It feels like every leader has been to some talking points boot camp where they’re incentivized to apply pressure to every part of their process - sort of a desperate attempt to justify the costs they’re incurring. I think we will look back at this and see how obviously short sighted it was.
LLMs mean less devs are needed, not no devs. even after serveral more decades, they'll need steering. I've seen agents stuck chasing one issue, when to me the issue was obvious, but I can see how the model would rule out the obvious easily and move on, but my instinct/experience tells me that's where I need to focus time on. This translates into costly token-waste. Secondly, it isn't simply "quality", the LLM might generate something that's good quality from its perspective, but it simply won't consider things unless it's explicitly told to in excruciating detail, and even then! understanding things from a simian point of view can't be perfected without that simian experience as part of its training. It can come very close but not quite.
Think of it this way, who needs engineering managers, project managers, scrum masters,etc.. if they're employable then surely actual devs that can tell what good architecture is vs bad, good code vs potentially bug code is are also employable.
But the number of devs needed, that demand will obviously decline dramatically. At the same time though, there are other careers that require programming and software dev as part of your skill set. Simply integrating LLM-enabled solutions into real world workflows is a new area that's very young and immature.
Let's not act like we're suddenly in some sort of post-scarcity utopia where all problems are solved by LLMs, where tech can solve problems, there is demand for those who can use technology to do so. However, I see a lot of people attacking the technology and resisting change a lot, and to those I suggest they look up every single technological revolution and see about the fate of such people.
This anonymous article is likely more FUD from the AI industry. "Just give up,you can't beat the machine. Please go quietly, we want to take your place and it's easier for everybody if you don't resist because you believe it's pointless"
'Maybe I should consider woodworking' - Fuck off.
This reads like a self thought ecommerce/small company employee finding its place in the tech world. These people were erased first, understandably. I am meeting more and more of the same type of people.
I had a friend in LA who was sure that CSS and HTML were enough for her to be a "Senior frontend developer". This year she moved to Tennessee and is trying to find a rich husband because she can't find a single job.
Shoemakers and horseback messengers complaining while Nike and FedEx deliver a million shoes or packages per day
We won't miss them
don't worry, soon there will be no "software engineering" careers anymore.
I've wondered about this a lot. I am brand new to software engineering, fully powered by AI coding. Traditional software engineers have to pivot hard or the are going to be left in the dust. The slow, methodical, take two days to put a change on a production site approach are over. I'm shipping exponentially faster than a co-worker who hasn't embraced AI yet.
My job as a staff engineer has turned into just reviewing slop farm vomit from offshore devs in Pakistan making pennies on the dollar given a slop code subscription and going wild.
I’ve lately just turned to having Claude do a quick /review, spot checking it, doing my own review and the. firing up some web agents to make the needed changes and just ignoring the back and forth because they don’t give a fuck anyway.
Just waiting for someone to notice and ask the obvious question at this point.
So blog with single post hyping LLMs. Oh and the domain name "human-in-the-loop". Call me suspicious.
If corporations really thought LLMs were a great cost-savings tool, then the obvious target for replacement are not the lower-paid staff, but the higher-paid staff - the ‘product managers and stakeholders’. That justifies token burn, replacing the 7- and 8-figure people, right?
But that’s not the real goal, is it? The goal is to inflate the stock value, take the cream off the top, and dump the whole business on the pension funds, maybe creating a too-big-to-fail scenario where the government steps in an bails out the industry as with the airlines during Covid.
This is why all the testimonials and narratives are so suspect - nobody knows what fraction of online posts were created simply to sell the narrative that LLMs are this incredible disruptive tool that will change the world, solely in order to create FOMO in the investor class.
In this particular case, I’d like to see links to samples of LLM created codebases for “PCI compliance, double-entry ledgers, escrows, reconciliation, payment lifecycles, bank transfer idempotency”. It should be easy to put an open-source LLM-generated version up on github, right? And if not, why not?
I started feeling like a factory worker well before LLMs. My reputation and network stopped mattering and it all came down to take this assessment and do this Leetcode to prove you are a good enough replaceable cog. I have about 15 more years before retirement and I doubt there is anything left to look forward to in my career.
Am I the only one that has noticed the massive increase in buggy software across almost every domain? Like, EVERYTHING has so many more bugs now. Things just break constantly. AI isn’t one shotting fixing bugs, it’s one shotting making hundreds of new ones every time it writes anything.
The glory days are over. In the future, one software engineer will be able to support multiple product areas much like how one HR team can support 1,000's of employees.
LLMs have made domain knowledge and reasoning "cheap"; it doesn't matter if the output is lower quality - look around you for countless examples of where cheap wins and "cheap" continues to improve.
Good luck out there; we will all need it.
I have no idea what you guys are up to, but it is just a job, it is just a role, it says nothing about you or who you are and it is not tied to your meaning. If you make it so and your perception is aligned with that, then you are not in control of what happens to you. What kind of slavery it is to give other people so much control over you is crazy
If the majority of the people have selected a direction you either opt in or opt out.
That's the hard truth.
Governments do dot care on our future, only on who pays them. This is the tragedy.
I think the author missed the forest for the trees - the domain knowledge is what allowed him to successfully use the AI because he instantly knew what was correct or not.
Constant use of AI will probably erode that knowledge over time just because of not practising it, but successful use in complex domain needs the domain knowledge to steer it away from icebergs or hallucination or model flaws.
move yourself to regenerative ag. take a look.
Of the posts I've seen by senior devs who assess recent events and end up roughly here:
> I'm still employed and I see myself employed for a foreseeable future. But I don't know what to think about the long-term ... Maybe I should consider transforming my woodworking hobby into a profession.
This one is notable for having all the clues pointing to why that's not the end-state this is headed toward, and yet... still not quite see it.
> I have no domain expertise that another Sr. engineer steering an LLM cannot match.
It's clear he's developed a significant competence in "steering an LLM" but the depth and value of that aren't apparent yet. After ~70 years, software development is now in the early stages of its first tectonic disruption. In the moment, these kinds of tech disruptions mostly appear to be displacing jobs but, historically, we understand the displacement is one part of a larger shift that's vertically compressing roles, functions and labor value. One steam shovel doesn't just displace dozens of pick-axe swinging diggers, it changes the roles, functions and competencies required across the entire supervision and management stack of "make tunnel through mountain" from the crew bosses and site managers to the tunnel engineers and business owners.
The author seems to be successfully navigating this shift but is still mid-disruption, so he and his management aren't yet able to see all the new competencies required or appreciate their value because it's all so new and still evolving. The rapid shock of agentic coding LLMs is especially disorienting because it's the first dramatic disruption in the field.
> review the code and steer the robot.
Historically, it's not surprising those few words are bearing so much weight and unappreciated value. Steam power was a similar shock to every field which relied on earth-moving and shaping. The big machines were quickly deployed, but it took quite a while for all the disruptions to both new and existing roles, functions and necessary competencies to be understood and appropriately valued. I imagine some top pick-axe swingers who'd graduated to being crew bosses and site foremen ended up driving or directing early steam shovels. In the first months they probably had little appreciation for the tremendous amounts of tacit new knowledge and practical expertise they accrued while keeping the steel beasts working. They were too busy being both amazed at the sheer power and frustrated by the constant scalding burns, tip-overs, blown boilers, landslides (too much weight, too little support) and cave-ins (dug too much tunnel, too fast with too little scaffolding), etc.
A big difference in the analogy is the first 100,000 steam shovels weren't sold at ~1/10th their actual cost and simultaneously delivered to job sites worldwide in six months. Software engineering is also unlike earth-moving and tunnel digging, in that the full costs and consequences aren't as visible or immediate as cave-ins and avalanches. The prices of 'steel beasts' are already going vertical with no end in sight and, over the next 18 months, I suspect "management" is about to gain a more viscerally accurate appreciation of the catastrophic costs of digging 'too much tunnel, too fast' absent the close supervision of highly skilled experts in directing all that newfound power constructively and not destructively. Between the skyrocketing full cost of operation and the consequences of poorly managed, non-expert execution - we'll start to see the broad outlines of the new equilibrium take shape.
In the steam era it over a decade for the ecosystem to understand how to even draw a new org chart accurately, label the boxes and appropriately value proven competency where it mattered. The faster the disruption, the longer it can take for all the pieces to rebalance and stabilize around a new equilibrium. Today, the author doesn't know all that he already knows and doesn't yet have the visibility to see how the new domain competencies he's rapidly accruing are creating a different kind of role that could be even higher value.
Just want to point out that code quality and architecture is actually eroded by codes 5.5. It’s over for this job I think.
I think we are all vulnerable and need to reassess what it is we bring.
Agents merely accelerate and equalize the playing field. And they cost money. We might be a dying breed, but we are the best operators of this technology. And if we want it, this is our moment.
Yes, get into wood working.
Isn't the solution to learn business skills?
My challenge is seeking good resources for the business skills. I'm doing sales for a passion project for the first time, and it's teaching me a lot. I'm just confused still on why it feels so hard and why I can't find an easier way.
Think of it like this
You’ve already faced this the entire time with… libraries on github.
If employers knew how much you can just use a new standard library, or ask you to “use React”, that’s a lot like asking you to use an LLM to speed things up. You also benefit from the collective wisdom of a lot of people. Do you write assembly or pixel shaders by hand?
I can't write what I really think because my name is attached to my account.
Let me just say AI is not nearly as good as the billions of dollars in marketing spend say.
We are months away from catastrophic bed shitting and the tech industry will pay the piper.
I'm not worried. You cannot hold a machine accountable and there's no way OpenAI, Anthropic, etc. are going to take on that kind of liability if some code resulted in a major outage or a lawsuit. Perhaps that's the signal I'd be looking for: so much confidence in the product that they put their money where their mouth is.
Besides, you can look at the websites/apps/software you use everyday and evaluate whether or not the agentic era has produced better results. Personally, there's still plenty of bugs and annoyances. Banks still using SMS 2FA, library breakages in minor version bumps, inconsistent UIs between web and mobile, etc.
If all that was a hurdle before... because humans, regulations, or something else... then surely these magical machines that can supposedly replace us and do it much faster would've handled it by now? And they wouldn't introduce more bugs[0], would they? ;)
So instead of a programmer, you become a software designer. I recently came across the idea of building fantasy for the player (in context of games), but now that I think more about it. Onlyfans, is just that. Advertising, Beauty products, novels, games, TV shows, and so on. You're really just creating / selling a fantasy for vast majority of people. Most people will never lose that 30 lbs, but you can sell them all kinds of products to fuel the fantasy of them losing that weight, being beautiful, rich, healthy and so on. So an LLM replacing the need for you to write every piece of code, is actually kind of freeing. You as a a former programmer, should embrace your new creative role. Writing code, at least for me was always slow and tedious. I just want to be able to express the ideas I have, so LLMs just make it possible to build things I never could otherwise.
Maybe just maybe here in HN we are in an echo chamber that is convincing us that there is a theoretical limit to how far the LLMs can make progress. It’s not unthinkable that LLMs will make better overall architectural decisions or follow the good practices better or understand the problem in bigger picture (more access to company/product context already makes a huge difference)
Lots of jobs have been automated away and careers based on those jobs faded away in history. Maybe in near future there won’t be a ton of opportunities for software engineers in the traditional form. I’m also embracing for that future.
There were people called calculators that did manual calculations in the past. There were people hand weaving all the fabric. There were people painting cars in the factory. All those jobs are gone for the most part.
We are sitting here portending there is going to be demand for software engineers managing those engineer robots but let’s be real. The demand for software is not increasing at the rate software engineering is becoming efficient using those robots. Some (many) of us have to find new careers.
> But now the market is shaping everyone into becoming a generalist.
This is interesting because in my field of VC everyone says generalists are dying.
The issue is that the people evaluating you don't know the difference between legit domain expertise and pure bullshit.
computers are made for automation. programmers were always working on automating things, making other things obsolete, and we have been killing jobs for decades. did you really think we would suddenly stop when it's your job? i'm happy this is happening, genuinely giddy
find another job but software engineering :) simple
These are the last days of software. Use the AI's and build cool shit NOW.
Yes this has been my experience as well.
It's crazy the crazed anti-AI people yelling with foam with their mouth that it's useless, meanwhile Claude for me at work oneshots complex bugs in a massive project with a 95% success rate. And the customer happiness survey has never been as good as it's now btw
"Except that nobody cares anymore."
:-(
People are missing the long-term horizon on this. Yes, definitely, you can automate most of your workflows as a software engineer with today's LLM frontier capabilities fully E2E. But many things are still super open: -First, cost is not a settled topic yet. We have no indication that automating everything E2E will be a cost-effective way of doing stuff. So the bare minimum is that you will need some expert designing the workflows in a token-efficient way. Worst-case scenario, tokens become super expensive and only certain parts of the job can be efficiently automated and many companies are not even able to afford tokens. -Second, the system you just "created" is just a static snapshot of today. Yeah it may work fully automated for 6 months, maybe a year. What then? Breaking changes? Updates? Re-designs? What if the quality slowly degrades until nothing ever works again? Who will fix that? There are so many unknowns that it is borderline irresponsible to make guesses on what can be automated sustainably long-term or not. Unless you are OpenAI's Codex team wasting a billion tokens a day on automating and self-improving everything, there is a high chance that everything you set up today is completely useless in a year. -Third, the core engineering workflow hasn't changed a single bit. People like stakeholders, product owners, PMs, etc. can come up with ideas and things to build but someone needs to take decisions on what gets built and what doesn't, balance out paying down technical debt vs. feature development, incorporate new domain knowledge into the system (Or would you expect your PM to be tweaking the prompts about a new regulation regarding GDPR or a completely new legal framework that changes the whole thing?) -Fourth, probably the most important one. If you think AI will soon get good enough to get self-improving and self-sustaining enough to replace full engineering departments E2E with no supervision then nothing else matters because we will all end up without a job and living on UBI (not only tech people). So why do you even care? If it happens it doesn't matter, and if it doesn't happen we just continue doing what we were doing until now. Why do you care?
The reason that I’m looking for an out is that it’s turned everyone I work with into imbeciles.
Nobody wants to think anymore. Coworkers are now just intermediaries for their LLMs. Talking to them is just talking to the LLM - sometimes directly copied and pasted, sometimes minimal effort to conceal what they’re doing. It is so disheartening.
And the sad part is, LLMs are incredible and can enable you to do much better work if you can stay in the loop, and stop focusing only on shipping speed. But from what I have observed, very few people care to do this. Who cares about substance when middle management thinks your productivity is 10x?