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

AI coding is gambling

notes.visaint.space

Post reply on HN
Print view View original

351–360 of 471 posts

Re: AI coding is gambling

#351

Post by IanCal »

input_sh wrote: post #332
IanCal wrote: post #219

Earlier quoted context omitted.

This is a pretty wild comparison in my opinion, it counts almost everything as gambling which means it has almost no use as a definition. The most obvious issue is it’d class working with humans as gambling. Fine if you want to make that as your definition but it seems unhelpful to the discussion.

You seem to have a fundamental issue understanding what the term deterministic even means. If you give the same trivial task to the same human five times in a row, let's say wash the dishes, your dishes are either gonna be equally clean or equally not clean enough every time. Hell, it might even get better over time by giving them feedback at the end of the task that it can learn from. If you run the same script five…

Deterministic doesn’t mean “generally pretty predictable, in broad strokes”.

> If you give the same trivial task to the same human five times in a row, let's say wash the dishes, your dishes are either gonna be equally clean or equally not clean enough every time.

Probably pretty similar but not quite the same. Sometimes they might drop a plate.

> If you ask the same question to the same LLM model five times in a row, are you getting the same result every time?

Probably pretty similar results. Sometimes they might mess up.

> It is different than a slot machine in a sense that you can influence those odds by asking "better", but that does not make it not gambling.

It rather can, we don’t call literally anything with a random element to the outcome gambling.

I’m probably gambling with my life if I pick a random stranger to operate on me. Am I gambling with my life if I take a considered look at the risk and reward and select a highly qualified surgeon?

Is it gambling to run a compiler given that bitflips can happen?

At what point does the word lose all meaning?

Re: AI coding is gambling

#352

Post by abcde666777 »

Bespoke suits are still a thing. Meaning where the highest quality is desired or valued, things still get handmade, and the rest of the time it happens on a factory line.

I suppose what's happening with software development is we're exploring where the line between the two is going to land. It's pretty clear that something like a simple and generic website can be reliably vibecoded, but on the other extreme I wouldn't expect the software for something like a space shuttle to be vibe coded due to the stringent safety requirements.

Re: AI coding is gambling

#353

Post by IanCal »

RhythmFox wrote: post #256
IanCal wrote: post #219

Earlier quoted context omitted.

This is a pretty wild comparison in my opinion, it counts almost everything as gambling which means it has almost no use as a definition. The most obvious issue is it’d class working with humans as gambling. Fine if you want to make that as your definition but it seems unhelpful to the discussion.

How does it 'count almost everything as gambling'? They just said 'non-deterministic' output is gambling-like, that is not 'almost everything'. Most computation that you use on a day-to-day basis (depending on how much you use AI now I suppose) is in all ways deterministic. Using probabilistic algorithms is not new, but it your point is not clicking...

Almost everything is non deterministic to some degree. Huge amounts of machine learning, most things that have some timing element to them in distributed systems, anything that might fail, anything involving humans, actual running computation given that bitflips can happen. At what point does labelling everything that has some random element “gambling” become pointless? At best it’ll be entirely different to how others use the term.

Re: AI coding is gambling

#354

Post by mh- »

mh- wrote: post #345
lcampbell wrote: post #309

Earlier quoted context omitted.

The reasoning generally isn't kept in the context, so after choosing the secret word in the first reasoning block, the LLM will have completely forgotten it in the second and subsequent requests. So, it technically didn't change the secret word so much as it was trying to infer what its own secret word might have been, based on your guesses.

Exactly. The following will work, assuming you're using a model and frontend that supports it: > Let's play hangman. Just pick a 3 letter word for now, I want to make sure this works. Pick the secret word up front and make sure to write the secret word and game state in a file that you'll have access to for the rest of the session, since you won't remember what word you chose otherwise. This was Opus 4.6 in Claude de…

On further experimentation, I prompted Opus 4.6 to make me a frontend artifact that used the Anthropic API, and I confirmed that it worked as expected.

Here is the only relevant part of the prompt it used when calling the API endpoint:

> - Track the conversation to remember your word and previous guesses

Re: AI coding is gambling

#355

Post by NewsaHackO »

datavalue wrote: post #325
NewsaHackO wrote: post #162

Earlier quoted context omitted.

I guess I agree with you, but I think the GP may have mispoke and meant he loves building software. It's sort of like the difference between knitting and making clothes. The GP likely loves making clothes on an abstract basis and realized that he won't have to knit anymore to do so. And he really never liked knitting in the first place, as it was just a means to an end.

It’s similar to the arrival of mechanized looms in the 19th century. My ancestors were weavers, and automation eventually replaced those jobs. I’ve spent 40 years working in IT as a programmer and am now nearing retirement, so I’ve been fortunate. To me it feels like programming as a skill may not have much time left. Probably how my ancestors felt.

Yeah, I was reading a little bit about knitting before my post and saw that in 1589, a person who invented a sort of prototype to the automated knitting machine in the UK had his patent application denied by the queen due to taking jobs away from hand knitters. I guess back then they had to be a little more protective because it would be a lot easier for civil unrest to lead to revolution and civil war in postfeudal UK than now.

Re: AI coding is gambling

#356

Post by pxc »

altruios wrote: post #322
_se wrote: post #277

Earlier quoted context omitted.

Are you implying that someone who prefers Eclipse is more likely to be a good software engineer than someone who prefers Emacs? If so, that is so hilariously backwards that I can't even begin to understand the types of experiences that you must've had. I am sure that you're objectively wrong if that is what you're saying.

I'm reading it as: those unwilling to try both and make an honest evaluation and instead have preconceived notions and bigotry tend to make bad programmers. That preferences are fine, but dogmatism should be avoided.

Nowadays most people try VSCode or JetBrains "by default" in school or at a first job. It's Emacs that's for explorers who actually try alternatives

Re: AI coding is gambling

#357

Post by paavope »

angrymouse wrote: post #296
Aurornis wrote: post #267

Earlier quoted context omitted.

I think the Product Manager title was (and still is) one of the most abused titles in tech. A great product manager is indispensable for setting product direction in a way that can't be accomplished by others doing it part-time or advocating for their own needs. I've worked with some truly great product managers. I've also worked with a lot of awful product managers. The product manager title is squishy enough that i…

What do you think the good ones do? And how do they set direction in a way that’s good compared to a bad one?

The good ones have original thoughts and can combine knowledge from different domains in nontrivial ways

Re: AI coding is gambling

#358

Post by bluefirebrand »

tossandthrow wrote: post #247
DrJokepu wrote: post #209

Earlier quoted context omitted.

Eh. You want a good mix of experience levels, what really matters is everyone should be talented. Less experienced colleagues are unburdened by yesterday’s lessons that may no longer be relevant today, they don’t have the same blind spots. Also, our profession is doomed if we won’t give less experienced colleagues a chance to shine.

Our profession is likely doomed not because we don't train people, but by the lack of demand

> I am never letting junior to midlevels into my team again

From a different one of your posts

So you're the one dooming the profession. Nice work, thank you!

Re: AI coding is gambling

#359

Post by 1313ed01 »

mh- wrote: post #345
lcampbell wrote: post #309

Earlier quoted context omitted.

The reasoning generally isn't kept in the context, so after choosing the secret word in the first reasoning block, the LLM will have completely forgotten it in the second and subsequent requests. So, it technically didn't change the secret word so much as it was trying to infer what its own secret word might have been, based on your guesses.

Exactly. The following will work, assuming you're using a model and frontend that supports it: > Let's play hangman. Just pick a 3 letter word for now, I want to make sure this works. Pick the secret word up front and make sure to write the secret word and game state in a file that you'll have access to for the rest of the session, since you won't remember what word you chose otherwise. This was Opus 4.6 in Claude de…

What you can do is to instruct it to type out the word, in some language that you don't know at all, making it available in the context while also effectively hidden from you. Simpler than printing it to a file.

Re: AI coding is gambling

#360

Post by tossandthrow »

MeetingsBrowser wrote: post #328
tossandthrow wrote: post #245

Earlier quoted context omitted.

That is not correct. It is much easier to control quality of an Ai than of inexperienced developers.

I think we are talking past each other. > I am never letting junior to midlevels into my team again My point is, you control the experience level of the engineers on your team. The fact that you can say you won't let junior or midlevels on your team proves that. You do not have that level of control with LLMs. Anthropic and OpenAI are roughly the same quality at any given time. The rest are not useful.

Ah, so that is not entirely correct.

I can control LLMs through skills and other gateways.

There are still tasks that LLMs does not really carry out that well, where a proper senior is needed.

Butnthese tasks are quickly disappearing, especially while the code base is slowly being optimized for agentic engineering.

Post reply on HN