> LLMs, by definition, can’t make anything thoughtfully, because LLMs have no thoughts. ... apps that feel crafted, that show refinement, that feel intuitive—requires a human developer not only with taste, but with understanding.
Well said, thank you.
It saddens me that people can't understand this. Already there is the fallacy of "I can make apps with AI, therefore no one needs to code anymore," as if all coding was was making apps. But even the apps that people think are passable feel so fake, clunky and sludgy.
That was there before LLMs. I know too many people who just shoved together some UI frameworks and called it a day. The same happens now with the only difference that LLMs are doing the shoving and therefore these things pop up at a higher rate. And to be honest the shoving that LLMs do, produce, most of the time, a better outcome than the shoving of people who don't care.
Yes there is the 1% of companies / people who really care about the things that you quote and these will be there in the future and, with the help of AI / LLMs refine / accelerate their workflow.
LLMs will give responses based on data that already exists. There's plenty of data that allows an LLM to write working code - possibly even high-quality code! But there's not much data that says how to make a "great" app, I think because it's domain specific.
Consider:
What made Firefox great was the lived experience of people fed up with IE6. The Internet talked plenty about speed, tabs, and stability, but "speed, tabs, and stability" undersells much of what made Firefox stand out.
What made Ocarina of Time great was that it defined a genre. No one really knew yet whether the story, the graphics, the dungeon design, or the marketing was going to sell that game. You couldn't yet ask any other Miyamotos what had worked for a 3D adventure.
We still arguing over what made Tom Brady (the American football player) great, not being able to nail down if it was the length of his career, the number of times he won the Super Bowl, his individual game records, his coach, the rest of his team, or something else entirely.
I think, likewise, the criteria for a "great" app aren't known until the app is already great, and that gives an LLM no specific advantage.
The "problem" (if it is one) is the same as we had with previous craftsmen being replaced by mass-production. Since it doesn't "have" to be done by hand, it can be done quickly and cookie-cutter.
And the dirty secret is that 90% of "stuff" is crud already (literally CRUD or just a tool for a task, no thought needed).
Everyone loves those little apps that used to be so common on the Mac, lovingly created and amazingly excellent at some small task; but the vast majority of people use whatever tool barely works and move on with their life. AI is incredibly enabling there; even if all you ever use it for is automating away spreadsheets.
I mean... I don't think ai development is appropriate for all uses, but I do personally get quite a lot of value out of being able to just get a quick and entirely custom solution to a specific pain point.
Ex - I ripped the home launcher out of my Nvidia shields and replaced it with an entirely ai generated one that has child lock controls and doesn't show any ads.
The "human produced" product was already a piece of enshittified crap. The llm version didn't need to be a masterpiece, it just needed to more clearly match my needs and respect my time. The current ai launcher is notably faster, ad free, and much easier to use. It's not worse.
All the other options for child controls that I could buy we're much more complicated, and wanted subscription fees (30+/y).
---
So it's a balancing act. Pretending there's no value here at all is also incredibly disingenuous.
Should all professionals be doing ai only development? Obviously not.
Can this still be a useful and meaningful tool? Obviously yes.
If you (generic) haven't yet used AI to script away some annoyances that you know you could do but never got around to, you really need to try. It's an eye-opener for sure.
I think one counterpoint is that sometimes something tailored to you might outweigh something that is thoughtfully designed, but designed for everyone.
I do think people quickly vibe coding apps that they expect to get any real amount of market penetration is a flash in the pan that will fizzle out pretty quickly - after all, if it took you a few prompts to write it would take me the same to do that, so why exactly am I paying for your version?
But I've built a lot of apps with LLMs that I intend to use just for me - I don't think I'd fool anyone into thinking they're not coded almost entirely with an agent, but I don't really care - they match all of the quirks and ideosyncracies that I personally have for apps, so it feels much better than something that's one-size fits all, even if that one-size fits all thing had a lot of effort put into it.
I have a bunch of little utilities like that that work the way I want to work - a pomodoro timer, a task management app, a password manager - I'm sure 1password, or Todoist, or some time management app with a bunch of bells and whistles is better for the average user, but the stuff I vibe code is better for me, since I can tailor it to exactly what I want.
Right now, it's mostly tech people who are doing stuff like this, but I don't actually think these hyper personal apps require all that much technical skill to make, especially when they just run locally, so I could imagine a world where all sorts of people make things for themselves using prompts.
Is the explosion of this little personal utility apps worth the multi-trillion dollar investment?
Anthropic's published accounts will tell the story. I find the current implementation of agents on top of LLMs pretty fantastic as software tooling. I really hope they're here to stay. But that's not the story sustaining all this trillion dollar companies building, running, renting GPUs and "frontier model" development. The real question is: will this technology transform the economy (or at least the economics of software?).
So far I'm tending to see it as a continuation of the cloud play: incentivize ever more inefficient use of ever more expensive compute. Just look at the valuation of public "tech companies" over the past 10-15 years.
I'm excited to see how it all turns out, but I don't see another decade and a half of exponential growth in profits and stock prices. Feels like a historic peak right now.
Sure, I'm not making a comment on the viability of the industry in general. I'm just saying that at least with the pricing and ecosystem we have today, AI apps have a leg up on customization, which can beat out good design a lot of the time.
Nah. In the big picture, voice can only ever be a complementary interface, not a widespread substitute.
It's fundamentally some combination of slow, imprecise, and ambiguous for most practical tasks and more importantly it places heavy demands on both the environment and the capabilities of the user.
It can be profoundly satisfying or efficient when the stars align over those limitations, but in the scheme of what we want computer-operated tools to do, these scenarios ends up covering a very narrow set of use cases.
The ultimate problem with voice interfaces wasn't technological, but biological/sociological: most people are simply inarticulate and impatient. They don't have the capacity to express themselves well in speech and don't have the patience to trudge through misunderstanding and clarification cycles -- especially when they lack the reflective insight to understand why they're being misunderstood. That's why they're not public speakers or writers or counselors or programmers or such (and even many of them are poorly suited). So even while the most limiting technological barriers may have been cracked open now, we'll still have to produce tool controls that accommodate these inarticulate, impatient people.
A part of software UX design is instrumental interaction. I could replace my guitar playing with singing (or asking a model to play the right notes). But that won't give me the feeling of fingers against the strings, or of moving effortlessly across the fretboard. There's a fair amount of UX use that follows the same idea, where the software becomes an instrument of sorts: https://link.springer.com/rwe/10.1007/978-3-319-27648-9_23-1
Voice interfaces have been decent for a while now, and the majority of people still use a GUI for the vast majority of interactions. They have their place but they're not going to become the dominant mode of interaction for a bunch of non-technical reasons (primarily because people don't generally want to loudly announce their work at all times).
Adding a number you pulled out of your ass doesn't turn your subjective opinion into objective fact.
Show us the LLM created app that looks like it was designed by Jony Ive, I'll show you one that just rips off any of the myriad frameworks designed by humans to mimic his aesthetic.
Your assertion is that LLMs "create apps which are more crafted and have more refinement than 80% of the human made apps"
If, as you assert here, LLMs are functionally equivalent in their "creative" process to humans, and you admit (as you seem to) that LLMs simply "steal ideas" from humans (which is apparently what you believe human creativity is comprised of,) to what do you attribute the apparently superior taste and design skills that LLMs have over humans?
LLMs can now solve famous math problems which humans failed to for 100 years.
LLMs can now create animations as beautiful as those of top humans animators.
LLMs can now design apps better than the vast majority of programmers.
I don't know why, and frankly I don't care.
I'm sure you have a theory why LLMs are solving incredibly hard math problems, probably along the lines "they were not actually hard problems, LLMs don't think or discover new things, they just interpolate in the convex hull of training data, so probably you just had to look in the right place and humans didn't had time to do that due to the vast search space".
>I'm sure you have a theory why LLMs are solving incredibly hard problems, probably along the lines "they were not actually hard problems, you just had to look in the right place and humans didn't have time to do that".
Now you're moving the goalposts and arguing with strawmen. You should ask an LLM to try to form a cogent argument. "I don't know, and I don't care, I just know you're wrong" doesn't make for interesting conversation. I'll let you have the last word, you seem the type who insists upon it. Good day.
> I admit I have trouble reconciling “I don’t like using libraries I haven’t written” with “I love having robots write code for me”.
(speculating) The easier way out is to understand that the first approach didn't make much sense either (even at face value "wanted to understand his code inside and out" can be easily solved by reading/testing the library code), so it's easy to get hyped up by another fad, there isn't much inconsistency between those at the deeper level of personal fascination
For awards, the use of A.I. should be baked into the category descriptions/rules of eligibility up-front, and the organizers should be free to have a contest which is AI-free (in terms of contestants, I'd like to hope bots would always be banned from voting), or a contest which allows it, or separate contest categories for use and non-use.
For my part, I'm late to the game --- I used Google's LLM-enabled notebook to generate some sample DXF files early on, but I mostly hack away and copy-paste from Stackexchange and so forth --- flet.dev hitting 1.0 and offering an online studio did have me buying tokens and ultimately a subscription to make:
while waiting for my next month of tokens to show up by copy-pasting and in a couple of instances asking Copilot about errors.
Currently trying to work out how to get the previews in the first app to look right, maybe adding an optional toolpath preview to the latter, and how to get a larger and more complex version up-and-running with an architecture I can wrap my head around using a toolchain I feel comfortable with (rather than a browser app, my typical development chain is local driven by a .text file: https://github.com/WillAdams/gcodepreview using a Literate Programming system).
The weird line is for knowledge lookup, which is sadly becoming more and more the case to make up for the very intentional enshittification of 10's era search tools. Given the recent walking elephant(-ish thingy) in the room for games, would I be using Ai if I consulted to claude on an issue and saw (didn't directly use, but saw) a snippet of code? Many would say no, but the environmental angle would suggest I am still contributing to AI pollution and the ethical side would suggest I am ultimately relying on a fountain of unlicensed code to get that answer.
But if you think that; how would you begin to enforce such matters? There wouldn't be any trace of AI contributions in a repo. Some theoretical perfect AI detector wouldn't detect anything because I synthesized the scripted into my own mind and figured out an approach to fix my problem. But the "damage" was done, nonetheless. That's what this current situation is having me consider.
I think different parts of systems carry different levels of risk. Some time ago I proposed a model for assessing the risk in a granular way. Putting a value (CCD - Cost of Cognitive Debt) on individual parts may foster healthy discussions within teams. Elaborating on it here:
For wear reasons alone, I don't see myself typing stuff in full with the one set of hands I have, when I can also make the machine do it.
RSI (Repetitive strain injury) is no joke in our profession.
So that alone makes the moral 100% anti AI standpoint a bit weak and surprisingly ignorant, given that a lot of the anger is (rightfully) about ignorant tech bros bulldozing reality with slop.
Counterpoint, while LLMs can generate tons of stuff, you still need to type stuff to make it generate tons of stuff, so you're still typing. Just more output per typed word, but you can still spend all day typing.
RSI doesn't seem to be nearly as big an issue anymore as it was in e.g. the 2000s though, not sure why exactly. Speaking for myself, physical exercise and a flat keyboard resolved it.
Speech to text improvements combined with the massive improvement in natural language understanding are the often unsung heroes of the current AI trend. Somehow, many people still seem entirely unaware of this workflow.
Heh, we really are reliving the same processes that prompted (pun not intended) the Luddites to smash textile machines 200 odd years ago. And looking at what happened with physical manufacturing of goods, the Luddites were completely correct about what would happen. See, the Luddites were not against technology, instead they were against the increasing social stratification that would happen due to the manner in which the machines were owned and operated. The textile machines were owned by a small elite of factory owners, who could use the new automation to produce textile in a more efficient manner than what could be achieved via a human artisan doing the same product. This was good for the factory owner as they could get the equivalent or greater output while hiring fewer people, leading to better margins. In some ways it turned out to be good for the consumer, as they could now get goods cheaper, and in many ways said goods would be at a more uniform quality, even if the quality itself might have been worse than what a master crafts(wo)man would have been able to produce. And that left the artisan, the worker, the producer. While many would have been able to pivot to using such machines or find other employment, many would have been left wayside after being made redundant by automation. And they, understandably, took offence to that, because they were deprived of a livelihood to support themselves and their families, which in turn lead to desperation, and as we know, desperate people can be very dangerous to those who they perceive as being the cause of their suffering. See the various strikes, revolutions etc. of the 19th and 20th centuries.
Now, a common argument against applying the same logic to today and the use of AI is that AI would be coming for all mental labour, not just specific fields like software, but at least in my view, this misses the point. IMO the point is that the current batch of techbro oligarchs lack the decorum and taste to be the stewards of this technology. That instead of driving this technology such that it'd benefit all of humanity and try to help those who would become redundant by the machines taking jobs away, for example by lobbying for universal basic income or some other types of safety nets, the techbros gleefully tell the public and the boards of various companies how AI can be used to reduce the need for human resources and put people out of jobs as the machines automate the various industries, just as happened 200 years ago. They seem to revel in the fact that people are losing their jobs as a result of these things, and they seem to not have learnt anything from the Industrial Revolution.
IMO every other discussion about things such as the quality of code produced by LLMs is frankly secondary to this argument of the ownership of LLMs being in the hands of a select few instead of the commons. The arguments here are just as relevant as they were back in 1848, just saying.
We're in learning mode between the optimists and the pessimists about how much.
I think it's easier to be an optimist and believe that no amount is too much - because if an AI is better than you at something...well... using it makes you more powerful. If there are subtleties that are wrong, you probably weren't in a position to realise what they were anyhow. If one can make money out of it and only the other guy will lose their job then ... so be it.
Hence the optimists will push us all over the cliff as with many situations in the past - lets just build another coal fired power station or spray everything with DDT or use xrays to measure your foot size in the shops so you don't have to remove your shoes - and here's a nice round thing you can light on fire and breathe on to relax! We've done a lot of deadly things in the past before we've eventually learned at the cost of millions of lives. People got wealthy on it and we don't seem to have asked for our money back.
The issue is whether the outcome is something that can be ignored or hidden - whether it affects a small enough number of people at once to cause everyone else to say "I'm alright jack." We have evolved responses in the past from unions to campaigns to stop smoking to netzero efforts but none of it was easy or quick and none of it is over either.
Reading stuff like this, I mentally recoil from the tone of self-righteous moral outrage. The ethical issues around AI don't seem any different than the rest of modern capitalism. But things like industrialised farming, manufacturing, fossil fuels, chemical production, etc. don't seem to generate anything close to the quasi-religious fervour that hating AI inspires in certain people.
It's difficult to escape the conclusion that the difference is that the harm from those other industries mostly accrues to the poor and the working class, whereas AI is threatening the livelihoods of rich, educated westerners who thought their social status insulated them from economic shocks like this.
No, they do generate that outrage, you've just become used to ignoring it.
Besides, outrage is mostly a response to change, or seeing something in the news (a major outrage driver). If there's no news item to hang it on the outrage just sort of sits there uselessly.
There's a lot of outrage to go round. An entire US administration devoted to maximising gross national outrage will do that.
The class commentary affects why you're hearing about it, not what the baseline level of outrage is.
Also, we are mostly too young to have experienced the outrage generated by industrial farming, industrialized manufacturing etc., while the outrage generated by AI is right here, right now.
> The ethical issues around AI don't seem any different than the rest of modern capitalism.
AI makes copyright and patents obsolete. There are no protections left for original nor derivative works; AI will hoover it all up and reproduce it without qualm and at scale. It must be trained, and it can only be trained on what already exists.
> But things like industrialised farming, manufacturing, fossil fuels, chemical production, etc. don't seem to generate anything close to the quasi-religious fervour that hating AI inspires in certain people.
A computer is a general-purpose machine. A tractor is not. AI is threatening every job that involves using a computer to accomplish it, or that could use computers and automation to accomplish it. Not to mention the promise of exponential growth; AI will continue to improve and in turn be used to improve itself, which a tractor cannot do.
These kinds of threats are what inspire such fears.
"A computer is a general-purpose machine. A tractor is not."
I mean, a tractor can't do everything, but I'd be hard pressed to find something in the farmers tool box that can get a wide ranges of tasks completed.
Pretty much all farm tasks require it: Drilling, discing, spraying, cutting...not to mention pulling wagons, goods to market, etc etc.
Exactly, the tractor was the general purpose machine for agriculture and all of the land management tasks that were tied up with that and modern people forget that agriculture used to be 60% of employment. If you broaden that to the motor then it really is a general purpose machine for all physical labour, which is something a computer is not capable of doing.
> AI makes copyright and patents obsolete. There are no protections left for original nor derivative works; AI will hoover it all up and reproduce it without qualm and at scale. It must be trained, and it can only be trained on what already exists.
if an AI produces direct plagiarized text from an existing work, that's copyright violation no different from if I simply wrote in my own book someone elses' words without attribution; the copyright holder is entitled to the same legal relief in either case.
if OTOH someone reads a lot of books and then goes off and writes their own book, and the influences of those various books is obvious but they didn't copy any sections verbatim, this is not a copyright issue, a critic might say "wow they really ripped off X and Y, what a terrible, derivative work". But no legal copyright violation, and this is incredibly common, normal, and ordinary, that people make movies/music/books/TV/whatever that is clearly derived from well known works that came before it. Doing the exact same process but also injecting one's own intent into this mixture of "derivation" is actually a feature, not a bug, of artistic creation - creators have to familiarize with the canon of work that came before them in order to produce new work. So if I said, "I want to create a computer program that makes art", of *course* it needs to *look at art*, either via ML training or encoded in the types of algorithms it uses, in order to attempt that.
What people are actually freaking out about is not the creation of cheap derivations, because bad, soulless art has always existed, but the fact that AI's can produce this usually mediocre-to-bad content at tremendous scales for very little money, crowding out human creation. As though there's a planet out there of conscious beings just like us, who have been watching human creation for centuries, and they've created their own creation economy, and one day we or them figure out how to give us humans access to it. Aside from the "we've made contact with extra terrestrial lifeforms" part of it, there'd be similar economic freakouts that humans only want to read books and watch movies made by those amazing alpha centaurian artists who are sending it to us for free.
I think the AI obsoleting copyrights is more of a problem for non-fiction, because nobody is gonna pay the primary source. So it's a big question how will remuneration work in the world where AI is effectively an enclosed commons.
> The ethical issues around AI don't seem any different than the rest of modern capitalism.
Let's fix that while we're at it too. I'm a bit tired of the "we already ruined this much. What's one more broken window" handwave. Let's start fixing windows instead.
>It's difficult to escape the conclusion that the difference is that the harm from those other industries mostly accrues to the poor and the working class, whereas AI is threatening the livelihoods of rich, educated westerners who thought their social status insulated them from economic shocks like this.
If you somehow assume that these people never cared about fossil fuels, manufacturing, unethical farming, etc. Yes. It can be easy to make that conclusion. Yet somehow, those have all been hot button issues for 30+ years at this point. Maybe they simply see an old pattern emerging to repeat yet again?
Why should they have been? Did computers consume much more energy than the bigger amount of people working much longer + all the energy wasted on mistakes than the computers?
Read the definition of "analogy", its core lies beyond your simple misunderstanding of a "false/true statement". If the commonalities your suppose exist in your analogy don't really exist, it's a false analogy regardless of the fact that a true analogy would still not be an "exact equality".
I did not mean it to be a 1:1 analogy, but rather a similar situation in history. My contention was not "this is precisely like the introduction of the spreadsheet." It was a tangentially related comment wondering whether there was the same level of angst among accountants when their field was suddenly impacted by a new technology.
"hn commenters" indeed. Instead of addressing the substance of the criticism of your analogy you ventured into irrelevant (and incorrect) semantic tangent.
Like you continue to do - I understand that it wasn't meant to be 1:1, but it doesn't impact the assessment that it's a very bad one because the things you think are common are not even close.
What sort of ego leads a person who doesn't use these tools to write articles about their capabilities like this?
> Anecdotal tales of discovering Claude has written four versions of the same function in different places, finding giant sections of code that never get called, and “bug-finding” agents that imagine and even create bugs are growing rampant.
Growing rampant? As in their getting worse? Because that's not at all my experience.
> The LLMs themselves don’t even “know” they’re working with code.
lol What does this even mean? What does "know" mean in this context? Models identify the code, the language its written in, the domain you're targeting, etc.
> LLMs, by definition, can’t make anything thoughtfully, because LLMs have no thoughts. ... apps that feel crafted, that show refinement, that feel intuitive—requires a human developer not only with taste, but with understanding.
Well said, thank you.
It saddens me that people can't understand this. Already there is the fallacy of "I can make apps with AI, therefore no one needs to code anymore," as if all coding was was making apps. But even the apps that people think are passable feel so fake, clunky and sludgy.
That was there before LLMs. I know too many people who just shoved together some UI frameworks and called it a day. The same happens now with the only difference that LLMs are doing the shoving and therefore these things pop up at a higher rate. And to be honest the shoving that LLMs do, produce, most of the time, a better outcome than the shoving of people who don't care.
Yes there is the 1% of companies / people who really care about the things that you quote and these will be there in the future and, with the help of AI / LLMs refine / accelerate their workflow.
LLMs will give responses based on data that already exists. There's plenty of data that allows an LLM to write working code - possibly even high-quality code! But there's not much data that says how to make a "great" app, I think because it's domain specific.
Consider:
What made Firefox great was the lived experience of people fed up with IE6. The Internet talked plenty about speed, tabs, and stability, but "speed, tabs, and stability" undersells much of what made Firefox stand out.
What made Ocarina of Time great was that it defined a genre. No one really knew yet whether the story, the graphics, the dungeon design, or the marketing was going to sell that game. You couldn't yet ask any other Miyamotos what had worked for a 3D adventure.
We still arguing over what made Tom Brady (the American football player) great, not being able to nail down if it was the length of his career, the number of times he won the Super Bowl, his individual game records, his coach, the rest of his team, or something else entirely.
I think, likewise, the criteria for a "great" app aren't known until the app is already great, and that gives an LLM no specific advantage.
The "problem" (if it is one) is the same as we had with previous craftsmen being replaced by mass-production. Since it doesn't "have" to be done by hand, it can be done quickly and cookie-cutter.
And the dirty secret is that 90% of "stuff" is crud already (literally CRUD or just a tool for a task, no thought needed).
Everyone loves those little apps that used to be so common on the Mac, lovingly created and amazingly excellent at some small task; but the vast majority of people use whatever tool barely works and move on with their life. AI is incredibly enabling there; even if all you ever use it for is automating away spreadsheets.
It seems like almost every argument defending the slide into agentic development is either:
1. This is just like the move to [X deterministic technology]. (False analogy)
2. You think it was good before? Who cares if it's a bit worse. (Corruption mentality doomerism)
I mean... I don't think ai development is appropriate for all uses, but I do personally get quite a lot of value out of being able to just get a quick and entirely custom solution to a specific pain point.
Ex - I ripped the home launcher out of my Nvidia shields and replaced it with an entirely ai generated one that has child lock controls and doesn't show any ads.
The "human produced" product was already a piece of enshittified crap. The llm version didn't need to be a masterpiece, it just needed to more clearly match my needs and respect my time. The current ai launcher is notably faster, ad free, and much easier to use. It's not worse.
All the other options for child controls that I could buy we're much more complicated, and wanted subscription fees (30+/y).
---
So it's a balancing act. Pretending there's no value here at all is also incredibly disingenuous.
Should all professionals be doing ai only development? Obviously not.
Can this still be a useful and meaningful tool? Obviously yes.
I think thats cause you're talking to software engineering.
There's lot of places where the cost of some simple program to do X is just to site-specific that it's cost per unit of automation is just too high.
I assume a lot of laid of software developers will find those jobs and AI will fills those voids.
Once those voids are filled, they'll need to justify their value prop, and SaaS are natural targets of high niche and low code responsibilities.
If you (generic) haven't yet used AI to script away some annoyances that you know you could do but never got around to, you really need to try. It's an eye-opener for sure.
I think one counterpoint is that sometimes something tailored to you might outweigh something that is thoughtfully designed, but designed for everyone.
I do think people quickly vibe coding apps that they expect to get any real amount of market penetration is a flash in the pan that will fizzle out pretty quickly - after all, if it took you a few prompts to write it would take me the same to do that, so why exactly am I paying for your version?
But I've built a lot of apps with LLMs that I intend to use just for me - I don't think I'd fool anyone into thinking they're not coded almost entirely with an agent, but I don't really care - they match all of the quirks and ideosyncracies that I personally have for apps, so it feels much better than something that's one-size fits all, even if that one-size fits all thing had a lot of effort put into it.
I have a bunch of little utilities like that that work the way I want to work - a pomodoro timer, a task management app, a password manager - I'm sure 1password, or Todoist, or some time management app with a bunch of bells and whistles is better for the average user, but the stuff I vibe code is better for me, since I can tailor it to exactly what I want.
Right now, it's mostly tech people who are doing stuff like this, but I don't actually think these hyper personal apps require all that much technical skill to make, especially when they just run locally, so I could imagine a world where all sorts of people make things for themselves using prompts.
The real question was never the utility of LLMs.
Is the explosion of this little personal utility apps worth the multi-trillion dollar investment?
Anthropic's published accounts will tell the story. I find the current implementation of agents on top of LLMs pretty fantastic as software tooling. I really hope they're here to stay. But that's not the story sustaining all this trillion dollar companies building, running, renting GPUs and "frontier model" development. The real question is: will this technology transform the economy (or at least the economics of software?).
So far I'm tending to see it as a continuation of the cloud play: incentivize ever more inefficient use of ever more expensive compute. Just look at the valuation of public "tech companies" over the past 10-15 years.
I'm excited to see how it all turns out, but I don't see another decade and a half of exponential growth in profits and stock prices. Feels like a historic peak right now.
Sure, I'm not making a comment on the viability of the industry in general. I'm just saying that at least with the pricing and ecosystem we have today, AI apps have a leg up on customization, which can beat out good design a lot of the time.
Funny thing is that people still use AI to write GUI apps, while in the future the entire app can often be replaced by a voice interface.
Nah. In the big picture, voice can only ever be a complementary interface, not a widespread substitute.
It's fundamentally some combination of slow, imprecise, and ambiguous for most practical tasks and more importantly it places heavy demands on both the environment and the capabilities of the user.
It can be profoundly satisfying or efficient when the stars align over those limitations, but in the scheme of what we want computer-operated tools to do, these scenarios ends up covering a very narrow set of use cases.
The ultimate problem with voice interfaces wasn't technological, but biological/sociological: most people are simply inarticulate and impatient. They don't have the capacity to express themselves well in speech and don't have the patience to trudge through misunderstanding and clarification cycles -- especially when they lack the reflective insight to understand why they're being misunderstood. That's why they're not public speakers or writers or counselors or programmers or such (and even many of them are poorly suited). So even while the most limiting technological barriers may have been cracked open now, we'll still have to produce tool controls that accommodate these inarticulate, impatient people.
A part of software UX design is instrumental interaction. I could replace my guitar playing with singing (or asking a model to play the right notes). But that won't give me the feeling of fingers against the strings, or of moving effortlessly across the fretboard. There's a fair amount of UX use that follows the same idea, where the software becomes an instrument of sorts: https://link.springer.com/rwe/10.1007/978-3-319-27648-9_23-1
In the future, the concept of music will be different. AI will produce music and read your brainwaves for feedback to find your brain's eigenmodes.
Voice interfaces have been decent for a while now, and the majority of people still use a GUI for the vast majority of interactions. They have their place but they're not going to become the dominant mode of interaction for a bunch of non-technical reasons (primarily because people don't generally want to loudly announce their work at all times).
That's only because there is still work left to do that AI hasn't taken over yet.
More and more people are just talking to Claude (now through a textbox, but soon through voice)
That sounds like hell.
[dead]
LLMs today create apps which are more crafted and have more refinement than 80% of the human made apps.
This argument is "a Ferrari is a better car than a Ford". What point is this exactly making? Who can afford having Jonathan Ive designing their app?
Adding a number you pulled out of your ass doesn't turn your subjective opinion into objective fact.
Show us the LLM created app that looks like it was designed by Jony Ive, I'll show you one that just rips off any of the myriad frameworks designed by humans to mimic his aesthetic.
Show me one well crafted human designed app, I'll show you the preceding apps it stole ideas from.
Your assertion is that LLMs "create apps which are more crafted and have more refinement than 80% of the human made apps"
If, as you assert here, LLMs are functionally equivalent in their "creative" process to humans, and you admit (as you seem to) that LLMs simply "steal ideas" from humans (which is apparently what you believe human creativity is comprised of,) to what do you attribute the apparently superior taste and design skills that LLMs have over humans?
LLMs can now solve famous math problems which humans failed to for 100 years.
LLMs can now create animations as beautiful as those of top humans animators.
LLMs can now design apps better than the vast majority of programmers.
I don't know why, and frankly I don't care.
I'm sure you have a theory why LLMs are solving incredibly hard math problems, probably along the lines "they were not actually hard problems, LLMs don't think or discover new things, they just interpolate in the convex hull of training data, so probably you just had to look in the right place and humans didn't had time to do that due to the vast search space".
>I'm sure you have a theory why LLMs are solving incredibly hard problems, probably along the lines "they were not actually hard problems, you just had to look in the right place and humans didn't have time to do that".
Now you're moving the goalposts and arguing with strawmen. You should ask an LLM to try to form a cogent argument. "I don't know, and I don't care, I just know you're wrong" doesn't make for interesting conversation. I'll let you have the last word, you seem the type who insists upon it. Good day.
> I admit I have trouble reconciling “I don’t like using libraries I haven’t written” with “I love having robots write code for me”.
(speculating) The easier way out is to understand that the first approach didn't make much sense either (even at face value "wanted to understand his code inside and out" can be easily solved by reading/testing the library code), so it's easy to get hyped up by another fad, there isn't much inconsistency between those at the deeper level of personal fascination
For awards, the use of A.I. should be baked into the category descriptions/rules of eligibility up-front, and the organizers should be free to have a contest which is AI-free (in terms of contestants, I'd like to hope bots would always be banned from voting), or a contest which allows it, or separate contest categories for use and non-use.
For my part, I'm late to the game --- I used Google's LLM-enabled notebook to generate some sample DXF files early on, but I mostly hack away and copy-paste from Stackexchange and so forth --- flet.dev hitting 1.0 and offering an online studio did have me buying tokens and ultimately a subscription to make:
https://studio.flet.dev/apps/bicQWc3Aul
but for a long while it wasn't running and I was out of tokens, so I coded up:
https://studio.flet.dev/apps/46tp8GPq9g
while waiting for my next month of tokens to show up by copy-pasting and in a couple of instances asking Copilot about errors.
Currently trying to work out how to get the previews in the first app to look right, maybe adding an optional toolpath preview to the latter, and how to get a larger and more complex version up-and-running with an architecture I can wrap my head around using a toolchain I feel comfortable with (rather than a browser app, my typical development chain is local driven by a .text file: https://github.com/WillAdams/gcodepreview using a Literate Programming system).
The weird line is for knowledge lookup, which is sadly becoming more and more the case to make up for the very intentional enshittification of 10's era search tools. Given the recent walking elephant(-ish thingy) in the room for games, would I be using Ai if I consulted to claude on an issue and saw (didn't directly use, but saw) a snippet of code? Many would say no, but the environmental angle would suggest I am still contributing to AI pollution and the ethical side would suggest I am ultimately relying on a fountain of unlicensed code to get that answer.
But if you think that; how would you begin to enforce such matters? There wouldn't be any trace of AI contributions in a repo. Some theoretical perfect AI detector wouldn't detect anything because I synthesized the scripted into my own mind and figured out an approach to fix my problem. But the "damage" was done, nonetheless. That's what this current situation is having me consider.
I think different parts of systems carry different levels of risk. Some time ago I proposed a model for assessing the risk in a granular way. Putting a value (CCD - Cost of Cognitive Debt) on individual parts may foster healthy discussions within teams. Elaborating on it here:
https://epiq.sh/blog/cost-of-cognitive-debt
Yes, 0.
For wear reasons alone, I don't see myself typing stuff in full with the one set of hands I have, when I can also make the machine do it.
RSI (Repetitive strain injury) is no joke in our profession.
So that alone makes the moral 100% anti AI standpoint a bit weak and surprisingly ignorant, given that a lot of the anger is (rightfully) about ignorant tech bros bulldozing reality with slop.
Counterpoint, while LLMs can generate tons of stuff, you still need to type stuff to make it generate tons of stuff, so you're still typing. Just more output per typed word, but you can still spend all day typing.
RSI doesn't seem to be nearly as big an issue anymore as it was in e.g. the 2000s though, not sure why exactly. Speaking for myself, physical exercise and a flat keyboard resolved it.
Try out voice mode. Its awesome.
Speech to text improvements combined with the massive improvement in natural language understanding are the often unsung heroes of the current AI trend. Somehow, many people still seem entirely unaware of this workflow.
Heh, we really are reliving the same processes that prompted (pun not intended) the Luddites to smash textile machines 200 odd years ago. And looking at what happened with physical manufacturing of goods, the Luddites were completely correct about what would happen. See, the Luddites were not against technology, instead they were against the increasing social stratification that would happen due to the manner in which the machines were owned and operated. The textile machines were owned by a small elite of factory owners, who could use the new automation to produce textile in a more efficient manner than what could be achieved via a human artisan doing the same product. This was good for the factory owner as they could get the equivalent or greater output while hiring fewer people, leading to better margins. In some ways it turned out to be good for the consumer, as they could now get goods cheaper, and in many ways said goods would be at a more uniform quality, even if the quality itself might have been worse than what a master crafts(wo)man would have been able to produce. And that left the artisan, the worker, the producer. While many would have been able to pivot to using such machines or find other employment, many would have been left wayside after being made redundant by automation. And they, understandably, took offence to that, because they were deprived of a livelihood to support themselves and their families, which in turn lead to desperation, and as we know, desperate people can be very dangerous to those who they perceive as being the cause of their suffering. See the various strikes, revolutions etc. of the 19th and 20th centuries.
Now, a common argument against applying the same logic to today and the use of AI is that AI would be coming for all mental labour, not just specific fields like software, but at least in my view, this misses the point. IMO the point is that the current batch of techbro oligarchs lack the decorum and taste to be the stewards of this technology. That instead of driving this technology such that it'd benefit all of humanity and try to help those who would become redundant by the machines taking jobs away, for example by lobbying for universal basic income or some other types of safety nets, the techbros gleefully tell the public and the boards of various companies how AI can be used to reduce the need for human resources and put people out of jobs as the machines automate the various industries, just as happened 200 years ago. They seem to revel in the fact that people are losing their jobs as a result of these things, and they seem to not have learnt anything from the Industrial Revolution.
IMO every other discussion about things such as the quality of code produced by LLMs is frankly secondary to this argument of the ownership of LLMs being in the hands of a select few instead of the commons. The arguments here are just as relevant as they were back in 1848, just saying.
We're in learning mode between the optimists and the pessimists about how much.
I think it's easier to be an optimist and believe that no amount is too much - because if an AI is better than you at something...well... using it makes you more powerful. If there are subtleties that are wrong, you probably weren't in a position to realise what they were anyhow. If one can make money out of it and only the other guy will lose their job then ... so be it.
Hence the optimists will push us all over the cliff as with many situations in the past - lets just build another coal fired power station or spray everything with DDT or use xrays to measure your foot size in the shops so you don't have to remove your shoes - and here's a nice round thing you can light on fire and breathe on to relax! We've done a lot of deadly things in the past before we've eventually learned at the cost of millions of lives. People got wealthy on it and we don't seem to have asked for our money back.
The issue is whether the outcome is something that can be ignored or hidden - whether it affects a small enough number of people at once to cause everyone else to say "I'm alright jack." We have evolved responses in the past from unions to campaigns to stop smoking to netzero efforts but none of it was easy or quick and none of it is over either.
Way more than what this cat is looking for.
Feel bad for people like this, seems like an incredible silly way to go about your life.
separately there's something Ironic about believing that ai is soulless and not real while also being a furry.
The amount that won't make me lose my job so I can keep providing for my family.
Reading stuff like this, I mentally recoil from the tone of self-righteous moral outrage. The ethical issues around AI don't seem any different than the rest of modern capitalism. But things like industrialised farming, manufacturing, fossil fuels, chemical production, etc. don't seem to generate anything close to the quasi-religious fervour that hating AI inspires in certain people.
It's difficult to escape the conclusion that the difference is that the harm from those other industries mostly accrues to the poor and the working class, whereas AI is threatening the livelihoods of rich, educated westerners who thought their social status insulated them from economic shocks like this.
No, they do generate that outrage, you've just become used to ignoring it.
Besides, outrage is mostly a response to change, or seeing something in the news (a major outrage driver). If there's no news item to hang it on the outrage just sort of sits there uselessly.
There's a lot of outrage to go round. An entire US administration devoted to maximising gross national outrage will do that.
The class commentary affects why you're hearing about it, not what the baseline level of outrage is.
Also, we are mostly too young to have experienced the outrage generated by industrial farming, industrialized manufacturing etc., while the outrage generated by AI is right here, right now.
> The ethical issues around AI don't seem any different than the rest of modern capitalism.
AI makes copyright and patents obsolete. There are no protections left for original nor derivative works; AI will hoover it all up and reproduce it without qualm and at scale. It must be trained, and it can only be trained on what already exists.
> But things like industrialised farming, manufacturing, fossil fuels, chemical production, etc. don't seem to generate anything close to the quasi-religious fervour that hating AI inspires in certain people.
A computer is a general-purpose machine. A tractor is not. AI is threatening every job that involves using a computer to accomplish it, or that could use computers and automation to accomplish it. Not to mention the promise of exponential growth; AI will continue to improve and in turn be used to improve itself, which a tractor cannot do.
These kinds of threats are what inspire such fears.
"A computer is a general-purpose machine. A tractor is not."
I mean, a tractor can't do everything, but I'd be hard pressed to find something in the farmers tool box that can get a wide ranges of tasks completed.
Pretty much all farm tasks require it: Drilling, discing, spraying, cutting...not to mention pulling wagons, goods to market, etc etc.
Exactly, the tractor was the general purpose machine for agriculture and all of the land management tasks that were tied up with that and modern people forget that agriculture used to be 60% of employment. If you broaden that to the motor then it really is a general purpose machine for all physical labour, which is something a computer is not capable of doing.
I don't think it makes patents obsolete. It makes it easier to infringe on one, perhaps, but you can still get sued if you do.
> AI makes copyright and patents obsolete. There are no protections left for original nor derivative works; AI will hoover it all up and reproduce it without qualm and at scale. It must be trained, and it can only be trained on what already exists.
if an AI produces direct plagiarized text from an existing work, that's copyright violation no different from if I simply wrote in my own book someone elses' words without attribution; the copyright holder is entitled to the same legal relief in either case.
if OTOH someone reads a lot of books and then goes off and writes their own book, and the influences of those various books is obvious but they didn't copy any sections verbatim, this is not a copyright issue, a critic might say "wow they really ripped off X and Y, what a terrible, derivative work". But no legal copyright violation, and this is incredibly common, normal, and ordinary, that people make movies/music/books/TV/whatever that is clearly derived from well known works that came before it. Doing the exact same process but also injecting one's own intent into this mixture of "derivation" is actually a feature, not a bug, of artistic creation - creators have to familiarize with the canon of work that came before them in order to produce new work. So if I said, "I want to create a computer program that makes art", of *course* it needs to *look at art*, either via ML training or encoded in the types of algorithms it uses, in order to attempt that.
What people are actually freaking out about is not the creation of cheap derivations, because bad, soulless art has always existed, but the fact that AI's can produce this usually mediocre-to-bad content at tremendous scales for very little money, crowding out human creation. As though there's a planet out there of conscious beings just like us, who have been watching human creation for centuries, and they've created their own creation economy, and one day we or them figure out how to give us humans access to it. Aside from the "we've made contact with extra terrestrial lifeforms" part of it, there'd be similar economic freakouts that humans only want to read books and watch movies made by those amazing alpha centaurian artists who are sending it to us for free.
I think the AI obsoleting copyrights is more of a problem for non-fiction, because nobody is gonna pay the primary source. So it's a big question how will remuneration work in the world where AI is effectively an enclosed commons.
> The ethical issues around AI don't seem any different than the rest of modern capitalism.
Let's fix that while we're at it too. I'm a bit tired of the "we already ruined this much. What's one more broken window" handwave. Let's start fixing windows instead.
>It's difficult to escape the conclusion that the difference is that the harm from those other industries mostly accrues to the poor and the working class, whereas AI is threatening the livelihoods of rich, educated westerners who thought their social status insulated them from economic shocks like this.
If you somehow assume that these people never cared about fossil fuels, manufacturing, unethical farming, etc. Yes. It can be easy to make that conclusion. Yet somehow, those have all been hot button issues for 30+ years at this point. Maybe they simply see an old pattern emerging to repeat yet again?
Can we just fast-forward to the part where the moralistic hand wringing is over, and we all accept that this is a part of software development now?
Were accountants this internally conflicted about the energy use of computers when they switched to spreadsheets?
Why should they have been? Did computers consume much more energy than the bigger amount of people working much longer + all the energy wasted on mistakes than the computers?
Spreadsheets were deterministic, reliable, hard-coded tools. This is a false analogy.
All analogies are "false"
But not all are false
An analogy is by definition a lossy comparison, not an exact equality.
Read the definition of "analogy", its core lies beyond your simple misunderstanding of a "false/true statement". If the commonalities your suppose exist in your analogy don't really exist, it's a false analogy regardless of the fact that a true analogy would still not be an "exact equality".
I did not mean it to be a 1:1 analogy, but rather a similar situation in history. My contention was not "this is precisely like the introduction of the spreadsheet." It was a tangentially related comment wondering whether there was the same level of angst among accountants when their field was suddenly impacted by a new technology.
hn commenters man
"hn commenters" indeed. Instead of addressing the substance of the criticism of your analogy you ventured into irrelevant (and incorrect) semantic tangent.
Like you continue to do - I understand that it wasn't meant to be 1:1, but it doesn't impact the assessment that it's a very bad one because the things you think are common are not even close.
> Were accountants this internally conflicted about the energy use of computers when they switched to spreadsheets?
they were! https://en.wikipedia.org/wiki/Productivity_paradox
spoiler alert, they went with computers after all
Again, can we fast-forward
What sort of ego leads a person who doesn't use these tools to write articles about their capabilities like this?
> Anecdotal tales of discovering Claude has written four versions of the same function in different places, finding giant sections of code that never get called, and “bug-finding” agents that imagine and even create bugs are growing rampant.
Growing rampant? As in their getting worse? Because that's not at all my experience.
> The LLMs themselves don’t even “know” they’re working with code.
lol What does this even mean? What does "know" mean in this context? Models identify the code, the language its written in, the domain you're targeting, etc.
[dead]
Human slop.