I haven't found one, like I mention in the article; I'll edit it if someone proves me wrong.
I'm starting to get a feel for a pattern - the books tend to be more expensive, and also take longer on average to deliver (a few weeks, instead of a few days). The latter would be normal for rare editions and some third-party sellers, but if I'm ordering a popular book and it takes longer than usual to deliver I can kinda smell the dead rat. But the only way to know for sure is to open the box in disappointment.
I had a friend over and we talked about the subject. She owns a Penguin stock copy of Martin Eden and upon checking my print-on-demand copy her first reaction was: "Yea, this looks like crap, but above all, the type is making me dizzy".
I only mention it in passing the article but I'm regretting not showing pictures of how bad the page typesetting can get - perhaps I'll revise it this week. There's a substantial qualitative jump from "this book looks like a cheap knock-off" to "reading this is giving me a headache".
And yes, while I don't have a clue about the printing process, the image of an inkjet printer has also come to mind on occasion!
1) Something happened during 2025 that made the models (or crucially, the wrapping terminal-based apps like Claude Code or Codex) much better. I only type in the terminal anymore.
2) The quality of the code is still quite often terrible. Quadruple-nested control flow abounds. Software architecture in rather small scopes is unsound. People say AI is “good at front end” but I see the worst kind of atrocities there (a few days ago Codex 5.3 tried to inject a massive HTML element with a CSS before hack, rather than proprerly refactoring markup)
Two forces feel true simultaneously but in permanent tension. I still cannot make out my mind and see the synthesis in the dialectic, where this is truly going, if we’re meaningfully moving forward or mostly moving in circles.
> People say AI is “good at front end” but I see the worst kind of atrocities there
It's commonly universal to say "AI is great in X", where one is not professional in X. It's because that's how AI is designed: to output tokens according to stats, not logic, not semantic, and not meaning: stats.
Reading discussions online and comparing them to my own experience makes me feel crazy, because I've found today's LLMs and agents to be seemingly good at everything except writing code. Including everything else in software engineering around code (debugging, reviewing, reading code, brainstorming architecture, etc.) as well as discussing various questions in the humanities and sciences where I'm a dilettante. But whenever I've asked them to generate any substantial amount of code, beyond a few lines to demonstrate usage of some API I'm unfamiliar with, the results have always been terrible and I end up either throwing it out or rewriting almost all of it myself and spending more time than if I'd just written it myself from the start.
It's occurred to me that maybe this just shows that I'm better at writing code and/or worse at everything else than I'd realized.
This matches my experience too. The models write code that would never pass a review normally. Mega functions, "copy and pasted" code with small changes, deep nested conditionals and loops. All the stuff we've spent a lot of time trying to minimise!
You could argue it's OK because a model can always fix it later. But the problem comes when there's subtle logic bugs and its basically impossible to understand. Or fixing the bug in one place doesn't fix it in the 10 other places almost the same code exists.
I strongly suspect that LLMs, like all technologies, are going to follow an S curve of capability. The question is where in that S curve we are right now.
I only say that because I'm a shit frontend dev. Honestly, I'm not that bad anymore, but I'm still shit, and the AI will probably generate better code than I will.
As long as humans are needed to review code, it sounds your role evolves toward prompting and reviewing.
Which is akin to driving a car - the motor vehicle itself doesn’t know where to go. It requires you to prompt via steering and braking etc, and then to review what is happening in response.
That’s not necessarily a bad thing - reviewing code ultimately matters most. As long as what is produced is more often than not correct and legible.. now this is a different issue for which there isn’t a consensus across software engineer’s.
I don't think that reviewing code is so important as reviewing results. Nobody is reviewing the IL or assembly code when they write in higher level languages. It's the end result that matters in most cases.
But we don't evolve IL or assembly code as the system evolves. We regenerate it from scratch every time.
It is therefore not important whether some intermediate version of that low-level code was completely impossible to understand.
It is not so with LLM-written high-level code. More often than not, it does need to be understood and maintained by someone or something.
These days, I mainly focus on two things in LLM code reviews:
1. Making sure unit tests have good coverage of expected behaviours.
2. Making sure the model is making sound architectural decisions, to avoid accumulating tech debt that'll need to be paid back later. It's very hard to check this with unit tests.
We get stuck reviewing the output assembly when it's broken, and that does happen from time to time. The reason that doesn't happen often is that generation of assembly follows strict rules, which people have tried their best to test. That's not the behavior we're going to get out of a LLM.
Yes, prompts aren't analogous to higher-level code, they're analogous to wizards or something like that which were always rightly viewed with suspicion.
> 1) Something happened during 2025 that made the models (or crucially, the wrapping terminal-based apps like Claude Code or Codex) much better. I only type in the terminal anymore.
I have heard say that the change was better context management and compression.
A lot of enhancements came on the model side which in many ways enabled context engineering.
200k and now 1M contexts. Better context management was enabled by improvements in structured outputs/tool calling at the model level. Also reasoning models really upped the game “plan” mode wouldn’t work well without them.
Did you read the post yourself? It doesn’t sound like it. It is composed of the title and three mystical-sounding quotes. How is one supposed to engage with this? Doing literary critique? A counter point to the statement “I don’t use LLMs” would probably count as valid engagement in any circumstance but especially in this one.
I did. The three quotes clearly express a shared sentiment for enjoyment of building and learning while doing so. That's certainly something one can engage with by providing a counterpoint. But just saying "that's not what I do" isn't one.
The original poster “expresses a shared sentiment” by posting three quotes, but the poster you replied to, who offers a fairly detailed account of the value LLMs bring to their daily work life, and how they feel about it, does not. OK.
The original post is a blog post that somebody put into their blog. Its purpose isn't (necesaarily) to engage into a discussion or even interact with anybody. It's the root of a discussion tree, if you will, a place to make a bold statement or just express a random thought.
In contrast, the post I replied to is a response, which by definition (and purpose of this forum) is meant to contribute to a discussion. It's an inner node of a discussion tree and thereby needs to engage with the presented argument.
So, this is an apples-vs-oranges situation, not a double-standards situation.
The irony of starting to claim that someone doesn’t engage with an “argument” (put forth by three quotes, and nothing else), and then ending up with this absolute word salad and an irrelevant metaphysical quip on the categories.
I build tech for Executive Search (headhunting) and I’ve recently written about this on my website.
Luckily in my niche the pressure to do this is not so high. Execs often have enough leverage to not have to put up with this kind of thing.
As others have commented, I am skeptical that this is any better than a form or similar. This could be a solution looking for a problem, or rather, relatedly, poorly allocated VC money looking to impress investors. Massive new entrants in the space like Jack and Jill are pushing this.
I guess there’s a vision where these interviewing agents truly become reactive and intelligent, so that they can both extract meaningful, deep insights about the candidate, while providing equally meaningful answers about the company and position. Color me skeptical, but not an outright denialist.
Regardless of the effectiveness for hiring companies, I think we will be seeing it for a long time. Even if it doesn’t produce meaningful improvements they will keep using it as long as it’s not too expensive, because the supplier and VC pipeline will press to keep using it.
I see some people are already doing OSS projects in this direction. I could be interested in exploring this and making a bot that really works on behalf of the interviewee. Agent-to-Agent communications may well be the future we are heading to regardless of our sensitivities to it, and I think the interviewee side of the market should and can get meaningful representation in this new world. Get in touch if you’d like to join forces.
I have to use LinkedIn to sell. I only occasionally look at the feed but I am ruthlessly muting or blocking anyone who is blatantly foisting their AI drivel on other humans. I’ve had enough of this shit.
I’ve been unfollowing people for a while and the issue is rarely from within my network anymore… the feed shows a lot of posts from AI foisters who I don’t even follow.
Can we stop softening the blow? This isn't "drafted with at least major AI help", it's just straight up AI slop writing. Let's call a spade a spade. I have yet to meet anyone claiming they "write with AI help but thoughts are my own" that had anything interesting to say. I don't particularly agree with a lot of Simon Willison's posts but his proofreading prompt should pretty much be the line on what constitutes acceptable AI use for writing.
Grammar check, typo check, calls you out on factual mistakes and missing links and that's it. I've used this prompt once or twice for my own blog posts and it does just what you expect. You just don't end up with writing like this post by having AI "assistance" - you end up with this type of post by asking Claude, probably the same Claude that found the vulnerability to begin with, to make the whole ass blog post. No human thought went into this. If it did, I strongly urge the authors to change their writing style asap.
"So we decided to point our autonomous offensive agent at it. No credentials. No insider knowledge. And no human-in-the-loop. Just a domain name and a dream."
Your reaction is worse than the article. There's no way you could know for sure what their writing process was, but that doesn't stop you from making overconfident claims.
That's the problem with AI writing in a nutshell. In a blind, relatively short comparison (similarly used for RLHF), AI writing has a florid, punchy quality that intuitively feels like high quality writing.
But then after you read the exact same structure a dozen times a day on the web, it becomes like nails on the chalkboard. It's a combination of "too much of a good thing" with little variation throughout a long piece of prose, and basic pattern recognition of AI output from a model coalescing to a consistent style that can be spotted as if 1-3 human ghost writers wrote 1/4 of the content on the web.
One thing I've learned recently is a lot guys (like here) have been out here reading each word of a given company's tech blog, closely parsing each sentence construction.. I really cant imagine being even concious of the prose for something like this. A corporate blog, to me, has some base level of banality to it. It's like reading a cereal box and getting angry at the lack of nuance.
Like who cares? Is there really some nostalgia for a time before this? When reading some press release from a cybersecurity company was akin to Joyce or Nabakov or whatever? (Maybe Hemingway...)
We really gotta be picking our battles here imo, and this doesn't feel like a high priority target. Let companies be the weird inhuman things that they are.
Read a novel! They are great, I promise. Then when you read other stuff, maybe you won't feel so angry?
I've picked up reading again over the last year or so! Maybe, if anything, that is why I feel so angry. Writing and reading are how we communicate thoughts and ideas between people, humans, at scale. A grand fantasy novel evokes a thirst for adventure, a romance evokes a yearning for true love.
What makes me angry, is to use the feelings we associate with this process and disingenuously pretend that there is a human that wants to tell me something, just for it to be generated drivel.
Don't get me wrong, I don't mind reading AI content, but it should read like this: "Our AI agent 'hacked' (found unexposed API endpoints) x or y company, we asked it to summarize and here's what it said:" - now I know I am about to read generated content, and I can decide myself if I want to engage with it or not. Do you ever notice how nobody that uses AI writing does this? If using AI to produce creative media, including art, music, videos, and writing, is so innocuous, why do all the "AI creatives" so desperately want to hide it from you? Because they don't want you to know that it's generated. Their literal goal is to pretend to have a deeper understanding, a better outlook, on a given topic, than they actually have. I think it is sad for them to feel the need to do this, and sad for me to have to use my limited lifespan discerning it. That is why I am angry.
Anyway, there's no need to "closely parse each sentence construction" at all to identify this post is fully AI generated. It's about as clear as they come. If you have trouble identifying that, well, in the short term you're probably at a disadvantage. In the long term, if AI does ever become able to fully mimic human expression, it won't matter anyway, I guess.
ps: FWIW, I agree with you that of all places, some random AI company with an AI generated website reporting on their AI pentesting with AI is the least surprising thing - the entire company is slop, and it's very easy to see that. My initial post was more of a projection at the dozens of posts I've read from personal blogs in recent weeks where I had to carefully decide if someone's writing that they publish under their own name actually contains original thought or not.
Ah well I guess you are on the right side of this either way! No need to even explain. It seems that people really really do care, and its wrong to say maybe its ok that they don't have to in this case. I guess I get it, I am generally more wrong the right anyway, and yes, at the very least, I am clearly in some way sub literate and uncritical as a reader, who can't tell the difference anyway. Not really the guy to be giving his opinion here. I will go find some slop to enjoy while the adults figure out the important stuff! Thanks for teaching me the lesson here.
A vibe? It’s completely obvious AI slop with no attempt to make it legible. They didn’t even prompt out the emdashes. For such a cool finding this is extremely disappointing.
Exactly the same on the other side though. If we believed that Dario Amodei or Sam Altman really knew how The Economy in its entirety worked, which is what they’re constantly pretending to do, then we should give them the central planning keys to the kingdom and declare communism tomorrow. I’m not being entirely facetious.
I drive a Cupra (a Spanish brand, a spinoff of SEAT, owned by Volkswagen) and when I was driving my car out of the dealership I had to tell the salesman I was not interested in signing up for the free app they were trying to foist me. He was dumbfounded. The model was selling like hotcakes yet I was the first customer that had outright refused to sign up for their shitty app. They pulled off the inevitable switcheroo, it now costs a monthly fee and I don’t want to think about what insidious things they can do with it.
Chances are the dealer and manufacturer can still get telemetry data through the cellular modem built into the car. You will need to remove it to be reasonably sure that data isn't extracted.
The issue is that data is also shared with third parties, such as insurance companies and possibly the government. I could not, in good conscience, drive a car that spies on me and can be effectively controlled remotely.
I also had to take this approach with my LG TV. The OS actually had its use for a while but Apple TV has become my driver. I no longer wish to consent to LG’s EULAs which are starting to look like the legal corpus of a small nation. I’m also not interested in their software updates. Internet privileges: revoked.
It’s concerning nonetheless as others are pointing out that in the current trajectory the TVs may soon refuse to display any content unless connected.
I’ve read reports of these devices turning on wifi and auto connecting to known public wifi networks. Seems we went from a generation of technologists dismissing Stallman as paranoid to one living in his nightmare and not being appropriately familiar with his work, issues of art vs artist aside.
Stallman has always been right. Hes a radical, but he was always been right. He is basically prescient with seeing how private software would be used. He was just so early that people thought he was a crazy radical, but now he seems to be stricken with a case of being Cassandra.
There hasn't been open public wifi networks near where I live for over a decade. It also seems increasingly rare for businesses to have them (they usually have an SSID and password posted somewhere). I don't think this is a thing.
But here's where it might go.
Verizon and other cell companies bundle streaming apps with their plans. It's really not a far leap for them to bundle a TV as well. Especially if TVs get really expensive due to whatever factors - get a 120" TV for just $30 extra on your bill over the next 5 years. And Verizon could contract with an OEM to make a Verizon-specific model, and put a 5G modem in it, and lock it to Verizon service. Verizon's just an example here, AT&T, T-Mobile could do the same.
Business idea: a signal-jamming cover for your 65-inch TV.
It looks like shit, is difficult to install, and costs an arm and a leg, but at least it prevents egregious privacy violations from your average chaebol or CCP-intervened corporation!
Why not just change your wifi password so that the TV can't connect again (after you've got your OTAs but I guess you could have loaded them on a USB stick to flash instead of wifi)
Who cares? Companies are using your need to have the latest and greatest against you. It's overt manipulation. I'd rather watch an old CRT or nothing at all than allow some company to forcefully show me ads.
> Companies are using your need to have the latest and greatest against you
This is a false dichotomy.
My love of cinema drives me to have certain features in my TV: 4k, OLED, HDR. My hatred of ads drives to me buy certain products to use my with TV: Apple TV.
You don't really get a lot of options anymore. When people around don't really care and just buy this junk because it's cheap, and they "need" a new TV ever three to four years, for some reason, then you get priced out of the market pretty quickly. Even if you look for TV, and yes I want a TV, not a monitor, without all this junk, there's not really any options available locally anymore. I believe my only option is the Thompson Easy TV, which is great, if I needed a 43" TV or lower.
Apparently I can attempt to import one from Romania, but that seems fairly complicated. Even sites that recommend dumb TVs just recommend SmartTVs that works well as a dumb TV.
It's ridiculous that going to a Target/Bestbuy/etc you cannot find any non-smart TVs generally. I have had several older models of non-smart tvs that suddenly stop working after a few years. It's disgusting
It's either because the non-ad-driven tvs cost more, resulting in too few sales to sustain (because no lifetime revenue from data sales) or the lifetime revenue from data sales is so profitable that companies take the risk on being undercut by a market entrant that will sell dumb tvs.
My guess is that the vast majority of people will trade data for a cheaper price point every time (my wife is certainly one of these people), so the market just can't support the volume of sales necessary to make the price point of dumb tvs competitive.
A smart TV just means one that can show you advertising and hoover up personal data. This is additional revenue. What company would sell you anything else?
We don't really have physical access to it - in the sense that on your desktop computer you can boot off a usb drive and reinstall the OS. There is no way you can boot your TV off external media. So you have to hack the existing OS while running it.
The way rooting working on a TV is that you run some javascript in the TV browser that targets some vulnerability in the browser/OS to run some code that then gives you a way in. Or if it has a USB port (to watch videos off a usb drive), you play a specifically crafted video that targets some vulnerability in the media players, to again install some program that then lets you do more serious changes to the OS.
The thing stopped being so needy when I neutered its internet access. Maybe it’s still exfiltrating data but at least it has stopped making me anxious that I may need to consult a civil rights lawyer every time I saw their EULA.
I'm starting to get a feel for a pattern - the books tend to be more expensive, and also take longer on average to deliver (a few weeks, instead of a few days). The latter would be normal for rare editions and some third-party sellers, but if I'm ordering a popular book and it takes longer than usual to deliver I can kinda smell the dead rat. But the only way to know for sure is to open the box in disappointment.
reply