We end up with some good threads on HN in both directions, but there’s a weight to early voices that can bend a posts’ comments pro or anti AI which isn’t necessarily reflective of the quality of said post.

And AI needs to avoid feeding itself synthetic data. Maybe this whole Anti AI frenzy is just an attempt to feed AI human content.

You can also exclude github repos posted here that contain ai attribution. Filters are updated daily to catch exceptions.

This is not an AI problem.

The internet is a bad place with lots of bad actors and no one says anything about malicious state actors who pollute the internet (they might be called xenophobic !)

AI merely is another tool for bad actors to be better bad actors, but that doesn't mean AI is the problem.

But you didnt grow up reading BYTE magazine.

I would totally understand your position if it was, say, crypto that you wanted to exclude. The whole of crypto finance is an edifice of scam built on some interesting but not very useful maths, and there's nothing to redeem it. There are no interesting tech discussions to be had there.

And sure, there are plenty of AI scam artists out there (most seem to have switched from Crypto experts to AI experts in the last couple of years).

But the underlying tech that is being used in AI is not only interesting, but also useful. I'm seeing people who have never coded before produce some cool apps. OK, they're not production-grade, but that's still new people doing new and interesting stuff with this tech. My own workflow is profoundly different from what it was a year ago. I've seen old problems that were really incredibly difficult to fix collapse completely using this tech. It's a useful tool if you're developing software. I suspect there are other areas of human endeavour where it will be useful too. I very much doubt it will replace all human work, or become sentient.

I think we need to separate out the AI business, which is its usual mess of scam, exaggeration, and buffoonery, and the actual AI tech, which is producing some really useful tools.

This is legitimate. There are many topics in computing and tech.

Not wanting to read about AI in every category is much the same like not wanting to see a C++ forum filled with threads about Angular.

To use your example; if the next version of the C++ language was switching to interpreted and compiled C++ was becoming an obsolete thing of the past, then I would expect every forum discussing C++ to be discussing this change. Trying to exclude that topic and still read interesting news about C++ would be very difficult, obviously, because everyone would be expressing their opinion on that change.

LLMs are the biggest change in the tech industry for ~50 years. It's not a separate tech (like C++ and Angular) it's a fundamental change in the way we develop software across the board. Yes you can still hand-code software (in the same way that in the example you could still compile C++) but the entire industry is moving in a different direction, and trying to ignore that seems quixotic.

Obviously, everyone is entitled to ignore whatever they want, so good luck to OP. I hope they find their source of non-AI tech news.

Yes it is.

> fundamental change in the way we develop software across the board

Whether you produce your code using "ed" on a hard copy printer TTY, or use an IDE, is off topic in a C++ forum.

The way you develop software is a specific topic: the way-you-develop-software topic. Someone not interested in that topic should be able to easily filter that out from their feed, like any other topic.

I wrote an timesheet tracking system for my A-level in 1986 on my BBC micro (the Atom's big brother). It took me more than half an hour, but it worked and did useful things. I'm sure it had bugs in it.

So I remember this period you're describing. Your software chief was a dick, though turning up 30 mins late on your first day would be a firing offence in any job back then.

We learned how to code when coding was very simple. There were no GUIs, no mice, all professional software was menu-driven. Screens had one resolution. None of these systems were networked, you only ever had to deal with one user. Security was never an issue, there was no external interface at all. Performance was usually not an issue, because everything was so simple. The OS was tiny and did almost nothing [0]. You could write software quickly because everything was much, much, simpler back then.

I'm seeing people who have never coded before write useful apps in 2026, against our modern massively more complex and intricate tech stack, using LLMs. No it's not perfect, but it's improving. There is absolutely zero chance that they could write an app themselves, or learn how to do that in less than a year, like we did back in 80's, because everything is so much more complex now.

[0] yes there were systems that did do all these things, but they weren't on the machines we're talking about here.

Just to nitpick, because this was my first computer. The Apple ][+ from 1982 featured an 8- bit 6502 processor capable of addressing max. 64 KB. If I remember correctly, the base model shipped with 16 KB and could be upgraded.

Because of the disaster of screen resolutions, mobile form factors and all the previous dishonesty inherent in the past 30 years of software development?

That must be really useful. The machines are not the problem. UI layout was solved 40 years ago on all windowed and non windowed platforms.

Mobile breakpoints are not special.

How about the browser as a "failed platform"? Seems to fit! Hardware is irrelevant.

Now, no. All that work will be done by an LLM. I'm afraid we don't get to play at being the returning heroes like those old COBOL dudes did.

Is there a level of abstraction where human involvement will always be necessary? If so where?

- scaling laws exist

- downstream perf trends also exist (epoch capability index)

- gpt4 to gpt5 leap in capabilities every 16-18 months

- actual adoption and retention and engagement numbers are out of this world

- RL with verifiable rewards will get you to super human performance even with poor sample efficiency

- zero evidence of a plateau

I don’t know the future I am just skeptically looking at the data.

What data are you looking at?

Also, frontier token prices have remained roughly constant:

3.5 sonnet: $3/$15 3.7 sonnet: $3/$15 Opus 4: $15/$75 (opus tier) opus 4.1: same Opus 4.5: $5/$25 Opus 4.6 (same) 4.7 (same) 4.8 (Same) Fable: $10/$50

So Fable is cheaper than Opus 4 was at launch.

One thing that has increased quite significantly? Spending and adoption.

This reminds me of that. The spec is the new high level language. Code is ASM. ASM is like CPU microcode.

By the 200xs they were gone. Interestingly, I would say what killed them in the early 2000s wasn't actually compilers, it was the interpreted languages. Others may disagree. Even if they were dog slow by comparison, scripting languages made some things so much easier to program that it didn't matter. And then it prompted static languages to up their game to try to match that. By the time that process played out, people writing only in assembler couldn't keep up anymore.

The company also had an AS400 with a collection of COBOL programmers. They were utterly scathing of the new toy language for doing toy things on PCs. There was no way that VB would ever be a "real" language or that anyone would do anything "real" with it.

And yeah, in terms of serious computing, that's probably true. But the industry leapt at the new tools and tooling, and COBOL faded to obscurity (though there are still AS400s out there, and some of the code they wrote is still managing vast swathes of our essential services).

And all of that was less of a revolution in the industry than the last 12 months have been.

I'm a little reluctant to go too triumphalist on the "adapt or be left behind" bandwagon, because I think that loses some very important nuances as well... but on the other hand, last week I set up a new project and the AI chewed through a lot of boring boilerplate and setup and configuration in about 30 minutes that would have taken me at least a tedious week just a year ago. Anyone who is leaving that on the table is going to find it hard to compete. I still had to go over it line-by-line, and make quite a few changes here and there, but it was also way, way faster to do that then to do it all from scratch.

My position has been to encourage the skeptics to give it an honest try (no deliberately picking something you pretty much know it can't do, or withholding info from it so it fails) but at the same time, hold on to the skepticism, it helps you use it properly. Excessive trust in the AI is definitely a problem right now.

At this point, this is categorically false. Your mental model is very out of date.

The results may not be ideal but I can't say I've had a single output so far this year that wasn't usable.

They absolutely 100% produce usable results for large classes of problems. Usually problems on the more "boring" side, but that's most code.

Even local ones are getting "usable." As an experiment I used OpenCode + Qwen coder to code a text based side scroller game in Rust. Left my laptop on and went to bed and in the morning I had a passable demo that looked like the old DOS game ZZT. This was with a 30b model at 4-bit quantization, so pretty tiny by LLM standards.

BTW LLMs seem to do well with Rust. I think it might actually be the guard rails. It lets the LLM iterate against itself. Dynamic languages don't give any compile time feedback, so it's harder to do that. This is something that's 100% the opposite of what I would have expected.

Crypto projects don't define their price, only their emission schedule. One doesn't buy without an understanding of the schedule. There are laws that are enforced against crypto rugpulls. Fundamentally there is no scam with the large crypto projects.

The worst part about it is like 80% of the conversation is motivated speech.

That's worse than the technology itself.

Do you mean it’s puffery or scammy?

By that standard then literally everything that is opinion based (ideological) is motivated reasoning

With LLMs, everybody talks them up because they are invested in Nvidia or have other exposure (which is almost everybody considering with all the datacenter building if this industry tanks US economy is also), or because they are invested in having jobs.

specifically those who like me are very excited about the future and have been excited about every step of the way since they got started?

In my case this is all going fantastically as planned and almost perfectly on the Kurzweil timeline. so as a 40+ person who has been wanting an Oracle in my pocket forever, we are closer to that then we have ever been. why the fuck wouldn’t I be excited about that? there was nothing like this in the 1980s and that’s a fact.

The concept of the Dick Tracy watch was science fiction when I was a kid and my kids all have Dick Tracy watches casually. That’s unbelievable!

The technology is amazing. Our society is the problem. People are attributing to AI what are ultimately society problems, much like the nuclear debate a century ago.

Also, I notice that with age people are less anti LLMs. I have to wonder if it has to do with the fact that we have less time left and so we care less about long term repercussions.

It's not that AI is so great, but that we really are in a different era defined by creative constipation and politics.

What will unclog the jam is making something and sharing it. I don't think anyone cares if you use AI to create some of it as long as it's not slop or too derivative. Things will recover, but it will never be the same.

Great, now we need AI to exclude AI related news /s