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

The copilot delusion

deplet.ing

Post reply on HN
Print view View original

101–110 of 214 posts

Re: The copilot delusion

#101

Post by lunarcave »

The things that's most often missed in these discussions that "writing code" is the end artefact. It doesn't take into account the endless tradeoffs made in producing the said artefact - the journey to get there.

Just try implementing a feature with a junior, in a mildly complex codebase and you'd catch all the unconscious tradeoffs that you're making as an experienced developer. AI has some concept of what these tradeoffs are, but that's mostly by observation.

AI _does_ help with writing code. Keyword there being - "help".

But thinking is the human's job. LLMs can't/don't "think". Thinking how to get the AI to produce the output you want is also your job. You'd think less and less if models get better.

Re: The copilot delusion

#102

Post by sgarland »

seanmcdirmid wrote: post #95
anon7000 wrote: post #85

Earlier quoted context omitted.

Yeah, and the article talks about those ways in which AI is useful. Overall, the author doesn’t have a problem with experts using AI to help them. The main argument is that we’re calling AI a copilot, and many newbies may be trusting it or leaning on it too much, when in reality, it’s still a shitty coworker half the time. Real copilots are actually your peers and experts at what they do. > Now? We’re building a worl…

> Btw, I say this as someone who does use AI coding assistants. I don’t think it’s all bad or all good. But we can’t just wave away the downsides just because it’s useful Isn't this just the rehashed argument against interactive terminals in the 60s/70s (no longer need to think very carefully about what you enter into your punch cards!), debuggers (no longer spending time looking carefully at code to find bugs), Inte…

Interactive terminals didn’t write code for you, and also unlocked entirely new paradigms of programs. Debuggers, if anything, enabled deeper understanding. Intellisense is in fact a plague and should not exist. Stack Overflow, when abused, is nearly as bad as AI.

Re: The copilot delusion

#103

Post by yencabulator »

tptacek wrote: post #44

The real horror isn’t that AI will take our jobs—it’s that it will let people in who never wanted the job to begin with. Gross. Also: you could have said this about the spreadsheet.

Meanwhile,

> 88% of the Excel spreadsheets have errors

https://www.cassotis.com/insights/88-of-the-excel-spreadshee...

How many companies mismanaged their finances because they had an enthusiastic spreadsheet user in charge? From that article, we know a country did.

Re: The copilot delusion

#104

Post by mountainriver »

sgarland wrote: post #79

> The real horror isn’t that AI will take our jobs—it’s that it will let people in who never wanted the job to begin with. I fully agree. This already happened with the explosion of DevOps bullshit, where people with no understanding of Linux got jobs by memorizing abstractions. “Stop gatekeeping,” they say. “Stop blowing up prod, and read docs” I fire back.

This is already happening and it’s a colossal mess

Re: The copilot delusion

#105

Post by joks »

vibeCoder007 wrote: post #22

Clean code is only good for non-profit organization. Your paid to solve problem as fast as possible, not to code.

For profit organizations usually care a hell of a lot about minimizing production issues, actually. What are you talking about?

Re: The copilot delusion

#106

Post by pron »

Jcampuzano2 wrote: post #33
boxed wrote: post #27

Earlier quoted context omitted.

Clean code is nice. Working code is a requirement. You missed the point. AI slop doesn't just fail on point 1. It fails on point 2.

I work in an enterprise company where we just recently got access to use cursor, before that copilot. We have literally stood up entire services built practically entirely with AI that are deployed right now and consumers are using. AI does work with competent people behind the wheel. People can't keep hiding behind saying that it always churns out code that doesn't work. We are way past those days. If you don't you…

When I've seen examples of such cases, my impression was that they could be improved by taking the AI out of the picture and using some "low code" solution.

Re: The copilot delusion

#107

Post by yencabulator »

Jcampuzano2 wrote: post #39
lamename wrote: post #35

Earlier quoted context omitted.

They will demand use of AI tools for "productivity" and then complain when there are bugs in prod without realizing the root cause.

They won't blame the AI as the root cause. They will blame you. This is why I mention you need to be competent enough to understand what is being generated or they will find someone else who does. There's no 2 ways around it. AI is here to stay.

The idea that you will actually really review all that vibecoded slop code comes across as very naive.

The real question is how many companies have to accidentally expose their databases, suffer business-ruining data losses, and have downtime they are utterly unable to recover from quickly before CxOs start adjusting their opinions?

Last time I saw a "Show HN" of someone showing off their vibecoded project, it leaked their OpenAI API key to all users. If that's how you want to run your business, go right ahead.

Re: The copilot delusion

#108

Post by mplanchard »

Jcampuzano2 wrote: post #54
lukan wrote: post #32

Earlier quoted context omitted.

I lost all my snobism about that years ago and I do just follow the paradigm, "does it work". But for some reasons, even with AI I ain't on another level. And those reasons are, it all collapses very quickly once the complexity reaches an medium amount. And if I want to rely on things and debug them - I cannot just have a pile of generated garbage, that works as long as the sun is shining. For isolated tasks it works…

Most replies here make the claim that all AI generated code is "garbage". And I can't help but think most of the people who say that do not actually use it in their day to day with the most recent models and actually give it good instructions/requirements. No, it is not always perfect. Yes you will have to manually edit some of the code it generates. But yes it can and will generate good code if you know how to use i…

It writes code well enough in simple contexts, sometimes. But that code is also easy to write, indeed often easier to write than to review. It struggles in more complex contexts and with more complex constraints. Unfortunately, it’s the latter case where I most often have any desire to reach for an aid, and it has failed so consistently and so often there that I have largely stopped trying.

It’s nice when you need to do something simple in an unfamiliar but simple context, though.

It seems though that a lot of the narrative here from its proponents is that we’re just not trying hard enough to get it to solve our problems. It’s like vimmers who won’t shut up about how it’s worth the weeks of cratered productivity in order to reach editing nirvana (I say this as one of them).

Like with any tool, the learning curve has to be justified by the results, but the calculation is further complicated by the fact that the AI tooling landscape changes completely every 3-6 months. Do I want to spend all that time getting good at it now? No. I’ll probably spend more time learning to use it when it’s either easier to get results that actually feel useful or when it stops changing so often.

Until then I’ll keep firing it up every once in a while to have it write some bash or try to get it to write a unit test.

Re: The copilot delusion

#109

Post by yencabulator »

bluefirebrand wrote: post #7
malfist wrote: post #4

I feel this in my bones. Every day I'm getting challenged by leadership that we're not using AI enough, told that I should halve my estimates because "we'll use AI", and being told that there's a new AI tool that I have to adopt because someone is tracking KPIs related to adoption and if our team doesn't adopt enough AI tools we're going to be fired to give more headcount to those that do. It's like the world has los…

> Management has an AI shaped hammer and they're hitting everything to see if it's a nail. I really think we need to figure out how to cut back on management so we can get back to the business of actually doing work

Clearly the answer is to replace them with AI.

Re: The copilot delusion

#110

Post by sumeno »

abletonlive wrote: post #73
MeetingsBrowser wrote: post #65

Earlier quoted context omitted.

Kind of brutal, but if LLMs drastically improved your productivity I think it speaks more to your baseline productivity than the power of LLMs.

What's more likely A) all of this money being funneled into tech to build out trillions of dollars worth of infrastructure, a month over month increasing user base buying subscriptions for these llm services, every company buying seats for LLM because of the value that it provides - these people are wrong B) yappers on hackernews that claim they derive no productivity boost out of llms while showing absolutely nothin…

Yeah, definitely A, same reason why all banking is done exclusively on the blockchain now and NFTs are the only way to get music
Post reply on HN