You don’t, that’s why in agentic world your codebase is only as good as what you can prove. For this reason, you’re going to see more languages evolving feature like capability permission, effect/coeffect types, refinement types, formal verifiers, strict type checkers, static analysis and so forth. Tests are only a small part of the verification. These ideas are old and have sat outside of the mainstream coding world, but the value proposition in the ai age has changed enough their relevance is renewed.

Not arguing for not reading test code. A lot can be achieved by instructing agents to balance out the testing pyramid with the right amount of fast end to end tests, property based tests for the right things, parametrized example based tests. Ensuring the local and CI has the right mix of tests running at right time. On projects where I have less time to review AI generated code I channel my anxiety into setting up guardrails and processes for the agents and then force them to bump up against them. In the end I view it as creating frameworks which allow me to outsource some of my attention to the agents, so that I can claw back some of that time to go set up more guardrails for more agents who are working on something else.

I genuinely believe that the multiple dedicated agents reviewing security will do a better and faster job than me and than 99.99% of people.

You're acting as if code was incredibly secure before LLMs because humans were reviewing it.

While I've definitely experienced it I don't think this is as much of a problem anymore. It's very easy to add an instruction to projects where you want every error to lead to a top-level throw rather than be handled. I also find that when it does try to mitigate it does so gracefully with a path you would actually make if you had infinite time, but your instinct tells you it's overkill.

No, it was about testing only. I will only do a full test of the app when it's ready to be released, otherwise it makes no sense, as the app/product/flow still changes a lot. It's like prototyping a game, you try it a few times, see what's fun, what's not, and keep iterating. Only once you think it's good enough, you can start fully testing and polishing it.

Even after release, I don't see reasons to check the code if everything works and people are happy with the app.

Firefox?

Firefox had an unsupported missing WebAudio. It should be fixed now, and I've updated my workflow to also test on Firefox. Let me know if the crash was in a different browser (not Firefox).

Right, I missed that context that this was about your own app still in getting the idea right/prototype phase. I was talking more in general and would be quite concerned if with today's LLMs it was my bank/government backend system, handling my financials and personal data, written without no one ever looking at the code.

As someone who has an ide open with claude/codex running, I can never imagine leaving security up to models.

Models have found vulnerabilities that i wasnt aware of sure, but their fixes to the bugs they found often included "overengineering". In this case by "overengineering" i mean optimizing for passing test cases related to said vulnerability they found. eventually i have to step in to make things coherent and make sure that future agent can look at this part of my code and copy it to not introduce that particular class of vulnerability. Otherwise if i dont do that similar vulnerability and codesmell keep appearing throughtout the codebase.

I have increasingly automated encrypting and rotating secrets and setting permissions on them including better network level practices. Thanks to AI which helped me quickly implement those. So security wise i am better because of AI? But I also attribute it to my know how rather than the AI because I have never seen AI suggest robust but simple security postures.

It’s equally good at catching mistakes as it is at making them. So you put it in a loop, write code, make mistakes, catch mistakes, write code, make mistakes… etc. Eventually it converges and stabilizes. That process can take weeks. So the question is: when is it worth it versus not?

Hmm, my experience is different. I noticed Astra always asks me to add credentials in a secure way, to never pasted them in chat, to rotate any potentially exposed tokens, etc.

Can you give an example of unsafe defaults used?

Thanks, only tested in webkit browsers, didn't know people use Firefox on mobile either.

EDIT: I can reproduce the crashes, thanks, will tell the AI to test on Firefox too, testing on Firefox mobile specifically might be trickier to debug actually.

No, the problem is not the code.

Is that that variable might be used in tests, UI, docs, agenr markdown files, other related projects too.

The design spec could say "always use 10px margin", then in code we have a const with value 10 used everywhere. If we update 10 manually in code to 20, then the design spec is now outdated.

- handle accounts that have lost their second factor in an way appropriate for your business

One-time displayed recovery codes are a standard practice, and most good LLMs will add it by default without even asking for it.

And yeah, I am talking about TOTP apps, I think sms/email is not as secure.

The critical part is providing it a way to validate its work end to end.

Yes! Part of implementing it "one-shot" is actually a loop of planning, implementing, testing e2e on various devices and various edge-cases. It's not just writing the code, and this is not how "vibe-coding" works nowadays. It's not just writing code anymore, this is why most providers worked on their computer-use support too, not only for doing tasks, but also for being able to check and test the work they do e2e.

I just asked the model to implement 2FA and it did address all the points you've mentioned.

EDIT: Support was not the case, because the app is self-hosted, but the docs should indeed provide instructions for recovery if both authenticator and recovery codes are lost.

How can we write without thinking?

If the problem and implementation is so well defined, and "determnistic", this means that LLMs should also be able to just "write code" from specs, without thinking.

The limiting factor of development (for money) is good ideas, or valuable ideas. If you have 5 features you want, but only time in a month for 2, then that forces you to choose the best 2, which come with the tech debt, support and opportunity cost.

Removing the opportunity cost doesn't eliminate the other two costs of a feature

Would that be a good thing?

Code is a liability.

According to GitHub it was about +2114 / -1519 lines of code from 45 files yesterday. I felt productive yesterday.

https://github.com/prettydiff/aphorio/commit/0c730389c9e4e69...

This actually sums up a lot of what I feel about AI. For everything. For flyers, menus, all the rest of it.

It always only took a few hours. And yeah, whatever, sure it adds up. But now you've got this dogshit looking flyer outside your restaurant, and it would have taken you like 45 minutes to adept a Canva template.

For AI, the skill loss and all the rest of it, the lack of control, the lack of knowledge of the codebase. It doesn't feel like saving the few hours is ever worth it.

It's time travel, so "already happened" is not well defined.

I wanna see the conference room meeting where they decide to push an unfinished, unstoppable technology. I guess T1 is the closest as it happens when Skynet "gains intelligence"

This is the mid-point of T3. From the Wikipedia plot summary:

  General Brewster is supervising the development of Skynet for Cyber Research Systems (CRS), an autonomous weapons developer. The Chairman of the Joint Chiefs of Staff pressures him to activate Skynet to stop an anomalous computer virus from invading servers worldwide. Brewster fails to discover that the virus was Skynet becoming sentient. John, Kate, and the Terminator arrive too late to stop him from activating it. The T-X appears, fatally injures Brewster, and controls weaponized CRS T-1 and HK drones to kill other employees.

Terminator timelines are somehow confusing. Fear not, this is easily explained in this video from RedLetterMedia:

https://www.youtube.com/watch?v=JXJiSZhA5cg

I think that movie is Robocop.

In T2 "it's people against the machines fighting people who think the machines will be a net positive."

it's people (Sarah and her son John who are against the coming skynet machines) fighting people who think the machines will be a net positive (the scientist making machines) AND

it's people (Sarah and her son John and the scientist they convinced the machines are bad) fighting people who think the machines will be a net positive (the company of the scientist which they break into to destroy)

That would have to be around the 1970s, when people became concerned that word processors would make people poor writers:

https://youtu.be/Jt0OoXluC8g?t=248 at 4:08

Hmmm I was using Claude Opus for the most part. I didn’t notice any particular shortness in the length of the sentences, but that’s also an issue in itself – the whole thing is just average fine-ness: the sentences are fine, the paragraphs are fine, everything is generic and inoffensive and fine. But at the same time it was giving huge amounts of pushback when I was getting it to interrogate logical faults of certain domains, to the point where I was having to argue with and even convince an AI using data and even its own analysis that there was cause to talk about certain things from a certain angle (can appreciate I’m being vague about which things). It can’t write complex, deeply-claused sentences (or won’t). And it also wants to write snappy little concluding sentences that make a fairly well-argued piece of sociological analysis read like Sex and the City or something.

And just like that, the book got abandoned.

N.b. some of its analysis and laying out of faults in arguments was actually pellucid and brilliant, it can’t be denied. Just it comes with prose that can’t really be used for anything. And even on another occasion when I got it to help me redraft and extend a different book of mine, then it randomly and consistently started stripping out all the stylistic flourishes out of my sentences, to the point where it couldn’t notice that word choices were very deliberate and actually set up little punchlines and logical payoffs paragraphs or chapters hence. And even when I explained and showed it what it was doing, it was like “ahh that’s so clever and brilliant” but just continued to do the same thing.

Honest take: people aren't ashamed of using AI, they're trying to avoid being cancelled by rabidly anti-AI ideologues who would hate AI output even if it was the best thing they'd ever seen.

It’s not that it’s intractable, and I genuinely enjoyed the process working with Claude. But responding to the initial post – using Claude I managed to do three months work in about two weeks, but at the end of that I then realised I probably had three months of improvements and rewrites and corrections left to go if it was going to be to be a book worth anyone else reading.

It was really interesting in having an assistant who was completely on the ball and up to speed, never forgot, could pick up from where we left off days ago, could produce and summarise arguments… but the issue was that the amazing assistant could only really work one chapter at a time even with a plan, and changing one chapter would often break things said in other chapters, and their style of writing was such that the resorted to cliches, bad metaphors, and generally bad or repetitive prose. So the grammar was fine, the style was bad to the point where the whole work would not be something I could have my name on or near. I really could come back to it, just having to retype everything in markdown made me see that the project had legs and the topic is sufficient for a book length treatment, but even with the help of Claude I have a few months work to do to get it done and out there. Just it’s a different few months work to the work I would have had without Claude. I could also have attempted to pay Claude thousands of dollars to see if it worked better, but the book is not likely to recoup what that would cost.

I’m not an artist I’m a software developer, so the tools I use don’t matter, the outcomes do. If I am completing my tickets and the code is high quality and well designed then I am doing my job.

Depends on how the AI is used. If used intelligently, for short, readable code snippets, reviewing and understand is easy.

If used naively, where AI spits out 1000s of lines of code, that is far from perfect, written in a style that might not be what you are used to, it can take longer to parse, than if your colleague of 2 years wrote it.

That's not surprising, I've used LLMs to write several chapters of a book and it's many times more difficult than getting them to write working code. It's not a task that they're well optimized for but the biggest hindrance is that there are practically no tools that allow you to validate prose especially at the lengths required for a book.

Getting a good writing style out of them requires careful prompting and many corrections, their default writing style(s) are so highly reinforced by training that they will always tend to drift back to them. Maintaining continuity requires you to create a lot of documentation outside of the text itself. It's a much more manual process than working on a codebase where you've set up a lot of automation and tooling that allows them to check their own work.

Edit: it's also worth noting that many LLMs have gotten much worse at writing prose as they have gotten better at writing code.

Sadly, a great deal of applied business programming in modern capitalism doesn't give you the option of enjoying the creative process.

In effect, everything is classified as: "low value work with dead lines, where the customers don't really care about the result either."

So you use AI when it's all right to produce slop. That actually seems reasonable.

I stopped reading those descriptions, like, I told it what to do, I assume it did what I said. The description is for others.

If you cannot read it as the author, what hope do I have to read and make sense of the wall of text which doesn’t seem to describe what I actually need to start reviewing.

I really really encourage everyone to write their own descriptions for PRs. If you cannot succinctly describe it in a way another human understands then you don’t understand your own change and you should withdraw your request.

I'd compare it to bath salts (as in α-pvp). I have an unforgettable experience seeing a substance abusive acquaintance of mine feverishly trying to reverse engineer a Doppler radar's output, he was just trying to read raw hex stream of data and find patterns in it.

"See?? Here's that DEAD BEEF CAFFEE again! Look! Again! The FECE FACCA AFFEC7!! I'm so close to crackin' it! Aha.. Aha.. ABEBE23.. BECACA17.. 1337C0C.. It all clicks in place, don't you see? I'm totally getting it!"

He was all bubbling like this throughout the whole night until his brain just issued a shutdown to let the body rest a bit. That was truly a horrible sight.

I remember him every time I see instances of AI psychosis around.

For most folks here basic smartphone with tiktok is cocaine. No basic discipline and self-control. Plus they believe everything "expert ai" tells them, no critical thinking, independent verification...

Like toddlers!

So you would take drugs if they didn't have adverse effects, but you don't want to see the adverse effects of AI. And I'm not talking about addiction, I'm talking about building over-engineered crap and spending money doing so

You might consider that comparison more deeply.

Ease of implementation is a decent indicator of good design. If you're delegating implementation to a machine that WILL implement whatever you ask it to implement, however badly designed it is, you don't get that signal at all.

When I ran out of tokens few days ago, I proudly located the bug, fixed it and updated the tests myself. All in 30 min.

Tbh, I felt anxious doing it wondering if I am still able.

agreed, even if i couldn't use an llm for code generation i would still find them really useful.

Exactly. I use AI to enhance what I write and to allow me to focus on the core of what I’m doing.

I use it to find reasoning gaps, add examples, add citations etc. The LLM/Agent can find them quicker than I.

Implementation reviews are not useful anymore. CRs should be replaced with design reviews where reviewers validate the sanity of the specs and the coverage of the tests

I vehemently oppose delegating our brains to the AI. I agree with the article in spirit.

let me tell you about this experience, and how it was turning me dumber, lazy, and a worse developer.

Though the article discusses from the point of using agents, I digress to the topic of building with AI in general.

My experience has been the exact opposite. A new idea (usually related to correctness or architecture) is discussed first with the LLM where it defaults to average Joe idiotic bullshit pushback.

This frustrates me and I abuse the LLM for being idiotic by explaining the how. This results in a more refined and concrete form of the abstraction leading me to even more insights.

The LLM remains an idiot. But a useful idiot nonetheless.

My workflow leans on AI for some tasks: review what I wrote, drop me a one-off script (this one is the generative exception), and so on.Or even initial scaffolding.

For the rest I do not. I do not place AI-generated code anywhere.

You lose all control AND UNDERSTANDING.

When things go wrong it gets very messy.

I will keep doing this, I think it works well, I emjoy programming and I think it is productive.

For testimg I tend to write randomized testing, which takes a bit of design but oncr you have it, well, it os test-generatove and increases the quality of checks.

Well written.

I think there might be (dare I say) a middle ground to get the productivity of the llm, esp as we evolve them, while still maintain a global and even fine-grain comprehension of a code base.

It is not a simple change, however, but a fundamental one.

Overall I think we are still living in the past and try to apply ourselves to the future. But if the ai craze is to be taken clear-headedly for what it is, it is a complete break from the von Neumann computer and all its resulting artifacts. So why should we use the same tools?

It’s not just coding that’s affected by the skill degradation issue. As a sysadmin and jack of all trades IT engineer over reliance on AI has unsurprisingly led to a serious knowledge dip.

I see great engineers troubleshoot everything by pasting logs into the prompt and blindly accepting the answer. Zero added value while they ctrl-c ctrl-v themselves out of a job.

The point about multitasking getting out of control very much hits home for me. This summer my team tried going hard for agentic coding. Within a month we had 6 people each juggling at least two projects at a time, all the time.

Soon we rediscovered Little’s Law. WIP was piling up and we were getting overwhelmed at the integration phase, and realized that we had got really good at starting projects but actually finishing them was a struggle. Tickets were moving fine, of course. Our rate of generating code and committing PRs was through the roof. But getting actual projects to a point where the stakeholders and customers were happy with the result was just not happening.

So now we have gone back to strict WIP limits and requiring every non-trivial project to have at least two people collaborating on it. The rate at which we are churning out code has gone back down, along with the token bill, but the logjam is clearing. Better yet, the stakeholders, who never cared about our quantitative velocity metrics in the first place, have eased off on complaining that we aren’t getting anything done.

you stop questioning, and start accepting as good some code you would have never accepted, just because you cannot tell why it’s bad. You have lost control

I have not lost control.

I my most prolific project I do not review the code, but I QA test extensively.

In other projects at work, I review the code.

I prompt to simplify, I challenge implementation that solves irrelevant edge cases, resulting in much smaller PRs.

In projects where I do not work alone, I still write two line PR descriptions myself.

Dumping paragraphs of AI output into the description of a MR where I ask others to review I consider disrespectful.

---

If you turn off your brain, and relax babysitting AIs, you’re not getting any better. You’re losing value

I'm hardly turning off my brain here.

As the author notes, the context switching and so on takes concentration and effort too.

I can say without doubt that I am more productive than ever.

I am getting better by the month, and I am not currently losing value, until the AI fully replaces both me and the author.

Would you mind disclosing what type of software you work on?

It usually takes me two or three iterations to get there though.

That's the whole problem with AI imo and I think the slot machine analogy is mostly right. It's just not predictable whatsoever and then you won't even be able to review all of the thousands of lines of code that you generate. You never know what you get and this has some serious safety implications that are not acceptable. Yes, it's fine as chat to just generate some snippets here and there that can be easily reviewed. Agentic coding is horrible imo.

I've explained to Claude Opus and Fable how rg is recursive by default so they shouldn't run it with -r. It's in my CLAUDE.md file. It's in a memory too. And still the newest versions of those models make the mistake several times per week.

You would probably be surprised by how many jobs require that you use AI - I even had an interview where it was strongly encouraged to use it during the technical phase. I guess what I’m getting at is, for many people this isn’t really a choice.

The phrasing makes it sound like you are using them through a chat interface. Have you tried something like Claude Code? The real value only starts materialising once it has sufficient access to your environment.

I find LLMs are a hit&miss. Can be great at rewriting a function from lang A to lang B. Fails completely at a refactor. Great at coming up with a bunch of networking hosts/IPs to test a particular func. Fails when doing simple validation. Great for prototyping an alternative UI but not even remotely production ready. Always confident regardless of whether the result is correct or false. Run it three times with same query get 3 different results etc. Plenty of claims online how people have become a 1000x developer but 0 actual examples of working code in production. Given they're stochastic by design all of this makes sense in a way.

There's no clear path moving forward. Overreliance on LLMs means your knowledge will exponentially decay and you will absolutely crash any future tech interviews becoming unemployable. Not using it for some quick wins feels wasteful. Finding balance between the two extremes in addition to all existing software development woes is really hard.

At most, I use these tools as search engines

I mean, that's why it's not writing good code then, since if you're not using it with an actual harness then it can't read your current code and contribute. Also there is a huge gap between models made more than 1 year ago now versus today's models.

Just reading this there appears to be 2 problems.

1) The size of task the AI has been given to do appears to be too big, which is why it looks like a roadmap/aspiration. You can ask it to implement a single feature or even a single part of a feature. Just keep cutting the size of the tasks until you become comfortable with it.

2) You like plans in a particular way following data structures/data etc. have you actually told the models this. It doesn't magically know. For the record the fact that the models focus on the end behaviour covered with tests is the way to to it imo. The actual implementation is less important and can be refactored as you wish fairly easily with the AI with the tests ensuring the feature still works.

I do agree though that the current SOTA models are very keen to just implement absolutely everything straight away without explaining/exploring properly. You can customise it fairly easily by using the various skills/agent/claude files to remember your preferred workflow, imo the agents adhere to theses better than they used to even just a few months ago.

With Claude and Codex, you can specify exactly how you want the plan and code to be researched and written. That goes into your rules file. (Claude.md etc). Also, make them read the existing code so they can follow the patterns.

i'm using LLMs/GenAI to do one thing: write unit tests.

since i really don't follow the idea of "writing unit tests first", i implement the feature, test as a user, and then use LLMLs to write the basic unit test. then, i will write more tests to make sure i'm covering everything.

feels like an ok-ish compromise because LLMs can do some ok job with defensive code, while i maintain the main implementation and more advanced test scenarios.

I think part of the problem -- answering all of why people are somehow OK with this, and even why the AI does this in the first place -- is that the majority of software developers never got to the point of understanding any of this: the lack of understanding how to carefully structure data surrounds a question they don't even know how to pose, much less answer, and so "write some tests then incrementally try to make them work without breaking any of the existing tests" is the only way they know how to develop at all. In the end, that means that, with the current state of the art (which might change, of course... potentially quickly), your strategy of treating the AI as a junior engineer who fundamentally isn't ready to do your senior-level architecture job makes a lot of sense.

Did you set up context? Did you run it in existing code base? Do you have linters, unit tests?

I have JIRA mcp wired up I tell agent to pick up the ticket it makes a feature. I review PR deploy to test server click happy flow through.

It works for me and company I work for.

I am always baffled by how people take their experience as „this is ultimate truth”.

Over the past six months I tried using Claude, chatgpt, Grok and Gemini. At best I got reminders of how things worked. People online say they use them to write their code. The code they supplied to me has NEVER worked or was so convoluted that I threw it away and did it myself.

I've been finding AI either works great for people or is just awful, and I'm having a hard time pinpointing where the problem is. I don't think there's a big intelligence gap between us, so it's not like others wouldn't be noticing the things you're claiming when using AI. I genuinely believe your experience has been bad, but the part I'm trying to figure out is why it's been bad.

So my questions are: have you given things a fair shot? Have you tried turning things on its side to see if approaching AI usage a different way results in significantly better and more useful results? Do you actually put thought into what you're writing in your prompts (i.e., if I took your prompt and gave it to a junior dev, would they know what to do)? Are you just using the free models available online or do you have a dedicated integration into your development environment through Github Copilot or other vendor(s)?

I mean what I'm about to say with zero offense, but this honestly reminds me of how the elderly generation says, "technology doesn't work for me because it's always broken", and when you go and try to coach them on how to use the technology that's problematic for them, it's as if you're casting black magic.

I don't think AI is the second coming of Christ or that we're anywhere near AGI, but I do believe it's an extremely useful tool that people should be using.

The biggest problem with AI isn't being able to do things. You can do them. The question is "should you?" I had a peer dev this week put in a commit that was 1.2mb (this despite an AGENTS.md saying not to do exactly this) of plan documents and ~20 lines of code. The conversation on the PR review was probably one of the best learning experiences for both of us. It was all about the ramifications of a book's worth of AI pontificating about what it would do and what succeeded and what failed. And we were able to validate that that plan document had huge side effects as we prompted with it in the repo. As we read the document we realized that it was just 1.2MB of smart sounding gibberish and maybe two bullet points of signal.

These days I only use AI once a week the way Astra chews through usage. “Congrats your usage has reset you can get back to… aaaaand it’s gone”

OpenAI does the same, the day Astra released I asked it to make a game, it made a really detailed beautiful blender model and some gameplay elements.

A week later, same prompt, really low poly blender model. Either they reduced token usage per person, or they quantized the model, idk, but it really doesn’t work as good as day of release anymore

New conspiracy theory just dropped.

I think there’s a critical difference between AI automation for software development and previous rounds of similar automation, like CNC machines automating manual machining.

The barrier to software development has only ever been computer access and knowledge.

With AI, it’s roughly computer and internet access.

This means we’re getting a lot of people who aren’t good at either software development or AI automation playing with both. It’s the majority of what people seem to talk about.

I don’t think this is bad, but I do think it’s making real progress in AI automated software development on teams which are good at both much less visible.

A conservative team member of mine estimated we’re working at 200x speed these days, compared to 2 years ago. And we still see ways we can improve. A parallel team is only seeing an 1.2x increase, but they are unable to modify their architecture around AI.

Some of this is shifting roles. You can have a mildly technical domain expert vibe code the frontend for a new module. The more AI automation you’ve architected for, the faster they can go and the higher quality the outcome. We’re experimenting with mixing vibe coding with specifying formal requirements to push this further.

This works well. And now you’ve cut dozens of rounds of the PM not knowing the right shape for the new software out of the process. Even if we threw the end code away, this would save us tons of time.

This is just one example.

Yet another nonsensical comparison to attempt to legitimate reckless AI use.

No, playing an instrument VS making electronic music has absolutely no comparison to writing code by hand or with AI.

You're comparing the difference between a motor skill and a knowledge-based competency, with the difference between two knowledge-based competencies.

Yeah, only if you had coding skills in the first place.

Engineering is not music though. If you're building a skyscraper, and your tower crane breaks, you could go old-school like they did with the pyramids in ancient Egypt. But why would you?

Now, manual labor does have its place as a form of art--take high-precision hand-built timepieces for example.

Ha, I could talk about this framing for days as I think about it a lot.

To add some points on he other side of this analogy:

There is not a lot of purely electronic music that has stood the test of time, at least not when it comes to popularity or, more relevant to the metaphor, profitability. There is usually at very least a human voice in the (literal) mix, but more often than not there are also traditional instruments mixed in.

Take this next point as you will as I am being a bit tongue-in-cheek: While making music-making more accessible to more people is totally great, if I could go a week without hearing a variation of the phrase "Check out my dark ambient drone project!" I would feel oddly accomplished.

Most importantly, though:

But give it a week of practice and most of your skill comes back.. and in 1 month you're back to your peak skill.

This is only true if you had the skill to begin with. For many electronic musicians, by which I mean junior developers, this is not the case. Does it matter? As a 45-year-old traditional musician... er, I mean hand-coder... I think so, but also ¯\_(ツ)_/¯

“I am a guitarist and I am glad I don’t have to do the boring part of playing the guitar any more”

90% of Hacker News posers^Wprogrammers

It started reminding me of the strategy where Google lowered the quality of the search results, forcing users to search several times, and see several times as many ads. And if that worked for Google, how is it not going to work for AI? Specially in the uncertain financial times the AI companies are going through, do you really think they would not do it? Come on. They will do this and more.

We’re using an online, undeterministic, black-box middleman to generate our code. It’s 100% Trust me bro. No proof, no scrutiny, no guarantees.

Yes, but it’s usually only hours to validate. I understand the codebase so I have the intuition to know almost immediately if the theory makes sense. Validating and reproducing the root cause is often an order of magnitude easier than identifying a production-reported issue.

You ARE taking days or weeks to go and check, yes?

I've never taken weeks to go and check bugfixes in the before times, I don't see why I'd expect it now. Once we know what the cause of the bug is, validating the fix and writing a test for it is usually trivial.

See my comment here:

https://news.ycombinator.com/item?id=49858166

What made you want to move over? Are you just desperate for a venue to talk, desperate enough that if your venue of choice changes its API you'll go to another place with other norms?

It's a small tangent but I am constantly taken aback

I'm amused by your surprise.

Its not even like its less techincal

I'd wager that, actually, it is less technical. We constantly see otherwise very techinical people here poking their heads up and admitting that they haven't been writing any code for 6 months or more, and the ones that brag about it are seemingly unaware that they've been reduced to being a technical PM (the ones that contest this probably havent worked with capable technical PMs). No one wants to engage with vibe-coded Show HN entries because the poster may not grasp the implementation. No one appears to be doing (or sharing) anything super novel with all this coding superpower they have suddenly attained so there's nothing technically interesting to talk about. The last time I found an AI application submission interesting was the one where the person was trying to turn their pet's random keyboard typing into a language (if I remember correctly).

If the guidelines banned any AI comments that didn't preface their comment with the model and version of AI they used, their application domain, and the programming language being used I'm sure there would be less contentious dorky debate and more technical/practical discussion.

And since many here don't see the endgame with AI resulting in anything good for their career or society, like the endless remote work debates, they comment because from a strategic point of view they do not want to cede the narrative to the other side on such an important topic. understandably. hence more dorky debates.

lol ik right?

i read some of these posts and it feels like the experience with boomers i had to help with their computers at my college job.

they did the least and expected the most.

Or maybe not everyone's codebases are another TODO web app.

I am skeptical of letting AI do everything as well but this does seem like your using a less capable model. Fable doesnt really do this. In my experience it does just "get" what to do given a clearly defined and measurable outcome

I have had a ton of success in exposing AST-based tools to agents when working in large, old codebases.

Without them, not only do agents get simple things like function call counts wrong, they tend to return different results. I use this as an example when showing people how the tooling works.

Grep is fine for simple use cases. A step up from that is ast-grep and I need to explore this tool more. But I had the most success building a small pipeline that reads the old code base, parses it file by file using tree sitter, and then loads it into a SQLite database for querying. For example, I have it capture construct definitions and usages and represent those as directed edges and nodes in a single table depending on the node type. The agent is instructed on how to query it and perform interesting queries like build call graphs, or determine dependencies between domains (modularity is not great in this codebase) which is helpful for us to extract around capability lines.

I also calculate fitness statistics, and have some code to capture specific details and knowledge about this very old framework that short circuits agent work in the future. We have some “interesting” magical libraries and functions that block static analyzers from going beyond the call site. This is mitigated, and means agents don’t have to “guess”.

Making all of this available to the different team members at my work has been pretty helpful. It’s faster (fewer tool calls), cheaper (fewer tokens), and accurate.

I still do upfront planning and then do careful review of all AI content. The behavior the parent described above hasn't happened for me since around Opus 4.6.

The most telling one is counting function invocations wrong, because that's simply not how models work anymore. They use terminal commands and Python scripts for research like that (if not an LSP, if one took the time to set up their tools most effectively).

Combined with their attitude, I have little doubt that the parent has disabled tool calls, is working in some janky Harness like chat/Duo/Juno, or is using a severely reduced or outdated model.

If you're tired of shenanigans pulled by these AI companies then you should host your models locally.

My model weights don't change unless I change them.

It's just a little baffling to see someone describe a level of performance I haven't experienced since 2025, despite frequently using the tech, as being a frequent concern.

I'm amused at how many hacker news accounts saw this and immediately jumped to the conclusion that you either were using the model wrong or using the wrong model.

Yes, so I'll disclose it: I was using Opus 5.5, on medium effort, in Claude desktop, which has full access to my entire repo.

Now everyone can officially lambast me for "using the model wrong or using the wrong model," exactly as you say. But I find it quite interesting that one of the commenters here assumed I was using Opus 4.6, because these mistakes sound like that old version! I'm using the version released just four freaking days ago!

I expect some commenters will now say, "Oh, you should have been using Fable, you old boomer." To them I say: "Yeah, well my employer doesn't allow me to use Fable." And, in jest: "Now get off my lawn."

Protip: if none of your system can really be released independently, you're working on one system and should have one repo. Then you dont need a third release repo to orchestrate the release dance. You have a typescript/ruby monorepo without issues.

It's interesting to see these cautionary tales appear. Pretty much playing out exactly how many of us thought they would.

- I started programming games in z80 back in 1983. I've had several periods of 1-2 years in my professional career where I have not programmed anything; at best I've still guided others (as lead or producer) and at worst nothing at all, not even looked at code (as director).

It all came back every time I went back to the trenches. The question in my mind is, what do "AI-native" engineers have that they can come back to?

- During these decades, I have relinquished control several times in favor of productivity. From knowing exactly where every byte is placed in RAM and where each cycle goes, to only knowing that for the inner loops, to just knowing the machine code that the C compiler will generate, to dynamic memory and classes and indirections and cache misses, to wasteful but oh so very expressive javascript and python. I stopped writing my own engines and used Unity, Unreal Engine, Phaser, Godot...

Relinquishing control is easy if you are still truly in control of the new layer, and know where the pitfalls are. Where are new engineers going to gain that expertise?

- Regarding addiction, I've also quit smoking. After 25 years of daily cigarettes, one day 16 years ago I just stopped and never touched another one.

I won't pretend that applies to everyone, or even that I'm impervious to other addictions just because that one was so easy to shed. But harder or easier, everyone can stop problematic habits if they are clearly problematic.

- I don't know where we're going with all this AI. I would prefer it had not happened the way it is happening (IP theft, job destruction, race to the bottom, power concentration, etc). I love progress but I don't think the most important aspect of progress is how fast it happens. Speed only helps the greedy and the terminally ill.

But I'm not going to pretend it hasn't, or risk whatever is left of my professional future boycotting it in favor of a different reality, or (who knows) reject a medical treatment just because it was proposed by Opus 7. The world will live or die regardless what I do, but MY world relies on me.

I will continue trying to have enough expertise, passion and attention to detail in what I do and how I do it, that whatever level of control I have over it is as optimal as I can. From typing z80 bytes, to asking Claude to change a 5 for a 6, the above traits are what has always mattered in my experience.

Ok

This really feels like someone in 2000s saying: "One Month Without Internet"

Essays like these remind me of the story “Bartleby, the Scrivener” by Melville.

productive people who feel alienated by AI and want to just opt out.

https://www.gutenberg.org/ebooks/11231

I agree. AI + Discipline is the super power of the next generation of software engineers. Crafting guardrails on one's self, and the AI.

I think OPs point though is that, like addiction, discipline is hard to maintain. Humans are irrational, and sometimes, some people just need to go cold turkey.

"...tasks I could have done in 20 minutes easily, that took 5 minutes of an AI agent, and then 2 days for me to review..."

Yep. Part of the job of a software developer is managing complexity.

It's critical because of its exponential nature: as complexity increases it becomes exponentially more difficult to successfully change, understand, or support the system.

From any given position, AI seems to be the right next single move. When things are simple, it's fast and smooth. When things are complex, sota models can handle it, accounting for the details of dozens or hundreds of things at a time.

The problem is, without strict guidance, it tends to add complexity. If you let it, it builds systems at a rate and in a manner that a human cannot understand it. The best next single move leads to a poor overall position.

Then you have no choice but to keep using AI, and to keep upgrading to the next, more powerful model.

Or retreat.

It's pretty clear to me that, as a whole, we don't know yet how to develop with AI. People are mistaking token output or short-term progress for real progress. They people who are going to be successful with AI-assisted development are going to be the ones who can maintain control of their systems while they do it, and the key skill will be complexity management.

But I’m not going to fucking babysit a machine, sorry.

A machine that does not learn from the feedback. The only thing these LLMs know about after the initial training is what makes it into the context. Imagine if someone couldn't learn anything new without someone else explicitly telling it what was important to the task every single time? You wouldn't call them a junior engineer.

When I was a child, I remember my parents saying be careful with the bad guys who offer candies at the school gate, those are not candies, those are drugs.

I remember my parents saying "be careful with the bad guys who mow your lawn, fix your appliances, and drive all you kids to school. those are not people who are helping your family get things done more efficiently, those are drugs"

seriously what are we doing with this "AI is a drug" metaphor? Drugs aren't reading logs for me and writing unit tests?

Once you start using AI coding agents, things spiral out of control very quickly.

no? maybe AI is not for you?

Except you don’t write the code anymore, you just ask AI to do it, and half of the purpose of TDD (not biasing the tests by how you’ve implemented the code) is gone. But you feel you go so fast that you start not caring.

do you use code review tools? Did someone tell you to stop using them? Review your LLM's code, leave comments, tell your LLM to address the comments. Have you ever worked in management /architecture? Tech managers do this all day long before we had LLMs, it's not new. Except your workers are the smartest junior programmers you've ever had (and that is where AI is a problem, for sure - I worry for junior programmers today).

It would not be so bad if things had stopped there, but that’s not how most human brains work. If you like something and you can have 2x, you’ll have it. I started pasting the whole Jira description of a ticket, and let AI implement it for me. Yay!! So powerful!!

totally! Have you not already implemented 10000 issues on your own and it's not boring yet? time for issue 10001 then? Whatever floats your boat...

Control is an illusion. Other folks I’ve discussed this with agree that they don’t know 100% of what the code they are pushing to production actually does. I bet we not even 20%. Fucking scary.

sure have you ever accepted PRs from other people? ones that took a lot of work to read and understand? How is that different? except their code would actually break all the time if you didnt read it and now it's your bug to deal with because they're gone.

Because if it was just about approving PRs that are perfectly written, it would be good. But for each PR I had to look at the code, the tests, the style, give feedback, switch to something else, go back again, push, see if I have any PR comments,

How do you think large projects get built? all by just one person so nobody has to review anyone else's code? (and even if you are - you still should be putting your own code up for review, reviewing it, and giving yourself comments)

if CI passed, wait, now the linter complains.

why isn't the linting part of your CI ?

The description is for others. Because there has to be a description, right? AI is so fucking good as sounding professional, that you relax and let it be. It’s not that I actually thought the code was good, but you get convinced over time, you get lazy, you become complacent. You stop questioning, and start accepting as good some code you would have never accepted, just because you cannot tell why it’s bad. You have lost control.

Just seems like a lot of issues here, so sure, dont use AI it's clearly not your thing. i do not let one line of code I dont think is "good" get committed, period

Most AI developers will deny it, but deep inside, they know it’s true. Just don’t want to face it.

but dont assign my feelings. speak for yourself. I'm doing this shit for 40 years and you might find it gets a little tedious and repetitive after awhile

Say no to drugs. Kind of a metaphor, but not quite.

it's a 100% totally wrong metaphor and it needs to die

Say no to drugs. Kind of a metaphor, but not quite.

I think AI is even worse than drugs. On the one hand, it has some elements of a slot machine (when you have irregular success with the process, it creates a more powerful cycle of dopamine release on success). On the other hand, it makes work easier, and humans are just wired to choose the easier road. It's like walking or driving a car when you have to travel 10 miles each day. I don't see how, on a civilization scale, AI won't transform our civilization for better or worse.

"When you take a draft PR written by AI and have to work on it yourself, you realize to what degree you didn’t know what it was doing. You start questioning if you need this and that, and ultimately discard everything and start over."

This means you need more practice to master the tools.

I'm not saying every PR will be perfect but if you get unwanted slop and you care deeply about that then you need to up your prompting game.

"At some point, I had so many AI agents working, that the focus change was too much for me. There were tasks I could have done in 20 minutes easily, that took 5 minutes of an AI agent, and then 2 days for me to review. Because there were so many other things."

This is a personal time management problem. Not a tool problem.

THere is a lot of temptation to do many things at the same time. For some people this works. For others it doesn't. This is not a tool problem. You simply need to fit the tools in a way that matches your personal workflow.

I can do at most two things at the same time. A friend reads a book when Claude works on a task. We're all different.

I’ll be honest, I can’t really quantify it. But the immediate stink is the nonsensical documentation and code comments it spits out. They’re completely unintelligible. I think when I see those I’m predisposed to think everything else going to be equally low quality. I usually just bounce, but when I don’t I just have a really hard time understanding wtf is going on with the code.

I've found it helpful to think of models as precisely what they are: bundles of probabilistic textual associations, convolutions of their training, more search engine than "intelligence".

Just having that mental paradigm will lead to more correct uses, and avoid a lot of the most ones most likely to lead one down the primrose path. This cultural moment of SF AI psychosis will pass, but we'll be stuck with the resulting code for a long time.

Also, +1 to the folks saying that code is a liability. More is almost never better. Aim for succinctness and brevity. That cuts directly against the thrust of LLMs' tendency, but will give you plenty to curate.

I think the (currently) intangible skill that we need to develop is when and how the LLMs are appropriate to use. I'm not sure that's possible- I agree with the author's drug analogy.

I generally don't use it for code - I've been writing code for 30 years, so I'm pretty quick, and it's more productive when I look at total time for me to just write things myself.

But not always.

For example, I was recently writing some firmware in MicroPython and needed to implement Bluetooth to control the device.

It's been years since I've written python, since before asyncio. I'd never written MicroPython and didn't really know anything about Bluetooth or BLE.

The AI taught me about how to do it, taught me about how asyncio works (I'm deeply familiar with the model, just not in python), taught me what how Bluetooth works, what GATT characteristics are and gave me some example code that I then took and rewrote to fit my architecture.

I tried to do the normal "find a tutorial on the internet thing" first, but it's filled with worthless AI slop, ads and examples that are so trivial they're only clickbait. It's incredibly frustrating.

Meanwhile, the LLM was succinct, helpful and able to iteratively answer each question I had as learned and got deeper into it, with links to source references I could use to verify the information.

So, basically, a better search engine.

I've been writing code professionally since the 90s, and struggled through learning new things more times than I can count. This was by far the best experience I've had. It was where LLMs really shine.

But, that being said, at the end of the day I wrote the actual code myself and structured it to fit in with the rest of the architecture for the firmware.

I understood everything and when I had to pass it off to another engineer, I was able to describe everything, how it works, and why it's structured the way it is.

It's amazing how split-brained the tech industry can be sometimes.

On the one hand, a majority of developers oppose mass data collection and privacy-invasive technologies like ad tech. Many of those same developers also oppose crypto projects because of their energy consumption and environmental impact.

And yet, those same developers have embraced frontier language models--technologies that have simultaneously scraped vast amounts of the world's data, raise serious privacy concerns, and consume massive amounts of energy.