Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI doesn’t reduce work, it intensifies it

simonwillison.net

Post reply on HN
Print view View original

211–220 of 329 posts

Re: AI doesn’t reduce work, it intensifies it

#211

Post by uyhsadf »

simonw wrote: post #196
tzury wrote: post #188

I would learn more about air combat by listening to a 12 minutes conversation with a jet fighter pilot, than I will from 3-day seminar by air force journalists. Tell me about what’s the LLM impact on your work, on account your work is not wiring about AI. Or if one wish for a more explicit noise filter: Don’t tell me what AI can do. Show me what you shipped with it that isn’t about AI.

> Show me what you shipped with it that isn’t about AI. From this weekend: https://github.com/simonw/sqlite-history-json and https://github.com/datasette/datasette-sqlite-history-json

[flagged]

Re: AI doesn’t reduce work, it intensifies it

#212

Post by sdf2erf »

thunky wrote: post #194
FieryTransition wrote: post #192

Earlier quoted context omitted.

And Unix was mainly made by two people, it's astounding that as I get older, even tech managers don't know "the mythical man month", and how software production generally scales.

Sorry but 99.999% of developers could not have built Unix. Or Winamp. Managers are crossing their fingers that devs they hire are no worse than average, and average isn't very good.

Thats because the distribution of developer quality and capability is skewed.

Re: AI doesn’t reduce work, it intensifies it

#213

Post by dcre »

cedws wrote: post #92
falloutx wrote: post #29

I am becoming more and more convinced that AI cant be used to make something better than what could have built before AI. You never needed 1000s of engineers to build software anyway, Winamp & VLC were build by less than four people. You only needed 1000s of people because the executive vision is always to add more useless junk into each product. And now with AI that might be even harder to avoid. This would mean the…

I’m finding that the code LLMs produce is just average. Not great, not terrible. Which makes sense, the model is basically a complex representation of the average of its training data right? If I want what I consider ‘good code’ I have to steer it. So I wouldn’t use LLMs to produce significant chunks of code for something I care about. And publishing vibe coded projects under my own GitHub user feels like it devalues…

People often describe the models as averaging their training data, but even for base models predicting the most likely next token this is imprecise and even misleading, because what is most likely is conditional on the input as well as what has been generated so far. So a strange input will produce a strange output — hardly an average or a reversion to the mean.

On top of that, the models people use have been heavily shaped by reinforcement learning, which rewards something quite different from the most likely next token. So I don’t think it’s clarifying to say “the model is basically a complex representation of the average of its training data.”

The average thing points to the real phenomenon of underspecified inputs leading to generic outputs, but modern agentic coding tools don’t have this problem the way the chat UIs did because they can take arbitrary input from the codebase.

Re: AI doesn’t reduce work, it intensifies it

#214

Post by uyhsadf »

simonw wrote: post #187
latexr wrote: post #177

Earlier quoted context omitted.

> One of my core beliefs is that "two things can be true at the same time". Which is, of course, true in some cases and false in others. But again, not what I’m talking about. > Got any examples that stood out to you of my inconsistency? Sorry, I don’t. You publish too often and obviously I’m not going to trawl through a sea of posts to find specific examples. I’m not trying to attack you. Again, my initial post was…

OK, I think I get it. Your primary argument here is that it's better to sit with ideas for a while before writing about them. My counter-argument is that's what I do... for my long form writing (aka "entries"). My link blog is faster reactions and has different standards - while I try to add value to everything I write there it's still a high volume of content where my goal is to be useful and accurate and interestin…

No his primary argument is that you are a fraudulent internet spammer masquerading as a software expert.

Re: AI doesn’t reduce work, it intensifies it

#215

Post by sdf2erf »

co_king_3 wrote: post #76
pousada wrote: post #43

Earlier quoted context omitted.

> percentage of good, well planned, consistent and coherent software is going to approach zero So everything stays exactly the same?

> So everything stays exactly the same? No, we get applications so hideously inefficient that your $3000 developer machine feels like it it's running a Pentium II with 256 MB of RAM. We get software that's as slow as it was 30 years ago, for no reason other than our own arrogance and apathy.

In a bizarre way all these new datacentre build outs may have 'fake demand' because of how inefficiently software gets produced.

Re: AI doesn’t reduce work, it intensifies it

#216

Post by ikr678 »

rybosworld wrote: post #206
cedws wrote: post #92

Earlier quoted context omitted.

I’m finding that the code LLMs produce is just average. Not great, not terrible. Which makes sense, the model is basically a complex representation of the average of its training data right? If I want what I consider ‘good code’ I have to steer it. So I wouldn’t use LLMs to produce significant chunks of code for something I care about. And publishing vibe coded projects under my own GitHub user feels like it devalues…

We've gone from "it's glorified auto-complete" to "the quality of working, end-to-end features, is average", in just ~2 years. I think it goes without saying that they will be writing "good code" in short time. I also wonder how much of this "I don't trust them yet" viewpoint is coming from people who are using agents the least. Is it rare that AI one-shots code that I would be willing to raise as a PR with my name o…

Is there a big enough dataset of 'good' code to train from though?

Re: AI doesn’t reduce work, it intensifies it

#217

Post by zozbot234 »

sdf2erf wrote: post #215
co_king_3 wrote: post #76

Earlier quoted context omitted.

> So everything stays exactly the same? No, we get applications so hideously inefficient that your $3000 developer machine feels like it it's running a Pentium II with 256 MB of RAM. We get software that's as slow as it was 30 years ago, for no reason other than our own arrogance and apathy.

In a bizarre way all these new datacentre build outs may have 'fake demand' because of how inefficiently software gets produced.

Much of the new datacenter capacity is for GPU-based training or inference, which are highly optimized already. But there's plenty of scope for optimizing other, more general workloads with some help from AI. DRAM has become far more expensive and a lot of DRAM use on the server is just plain waste that can be optimized away. Same for high-performance SSDs.

Re: AI doesn’t reduce work, it intensifies it

#218

Post by rybosworld »

ikr678 wrote: post #216
rybosworld wrote: post #206

Earlier quoted context omitted.

We've gone from "it's glorified auto-complete" to "the quality of working, end-to-end features, is average", in just ~2 years. I think it goes without saying that they will be writing "good code" in short time. I also wonder how much of this "I don't trust them yet" viewpoint is coming from people who are using agents the least. Is it rare that AI one-shots code that I would be willing to raise as a PR with my name o…

Is there a big enough dataset of 'good' code to train from though?

I (and lots of people) used to think the models would run out of training data and it would halt progress.

They did run out of human-authored training data (depending on who you ask), in 2024/2025. And they still improve.

Re: AI doesn’t reduce work, it intensifies it

#219

Post by rkomorn »

throwup238 wrote: post #198
rkomorn wrote: post #95

Earlier quoted context omitted.

I find it hard to disagree with this (sadly). I do feels things in general are more "snappy" at the OS level, but once you get into apps (local or web), things don't feel much better than 30 years ago. The two big exceptions for me are video and gaming. I wonder how people who work in CAD, media editing, or other "heavy" workloads etc, feel.

> I wonder how people who work in CAD, media editing, or other "heavy" workloads etc, feel. I’d let you know how I feel but I’m too busy restarting Solidworks after its third crash of the day. I pay thousands a year for the privilege.

Well that's sad to hear. Also kinda makes me glad I didn't wander down the path of learning Solidworks during the pandemic.

Re: AI doesn’t reduce work, it intensifies it

#220

Post by co_king_3 »

ikr678 wrote: post #216
rybosworld wrote: post #206

Earlier quoted context omitted.

We've gone from "it's glorified auto-complete" to "the quality of working, end-to-end features, is average", in just ~2 years. I think it goes without saying that they will be writing "good code" in short time. I also wonder how much of this "I don't trust them yet" viewpoint is coming from people who are using agents the least. Is it rare that AI one-shots code that I would be willing to raise as a PR with my name o…

Is there a big enough dataset of 'good' code to train from though?

AHEM

Let me repeat myself.

I think it goes without saying that they will be writing "good code" in short time.

Post reply on HN