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

I am leaving the AI party after one drink

lara-aigmueller.at

Post reply on HN
Print view View original

101–110 of 145 posts

Re: I am leaving the AI party after one drink

#101

Post by dd8601fn »

allenrb wrote: post #31

For what it’s worth (ie, absolutely nothing), I agree with her 100%. I didn’t get into this field in order to prompt an AI to take care of the details. I got into it because I love the details. I’m a strong performer on a good team at a company many people would want to work at… and I know the clock is ticking. Sooner or later, I will be too slow. I’m not going to claim that this is the wrong way to go. It’s obviousl…

Someone the other day was talking about there being two kinds of builders. One likes the details of doing, where the other likes the things they produce.

The idea was that one likes AI and the other naturally hates it.

I thought about that for a bit and decided that, like most things, if you’re any good at something the “hard way” you probably have some of both. Or at least I’m sure it’s true for me.

I LOVE that I can produce the things I want to create without spending months crafting lines of text. The “I know how to architect this, I know what a decent data model looks like, I have a good idea of where someone is likely to introduce security or scaling problems. I can pilot this plane and produce something GOOD.”

But, I really also HATE looking at the final product and forever measuring, in my head, how much of it is even mine. Which parts I haven’t thoroughly reviewed, or would have spent a week learning and didn’t, or maybe wouldn’t have accomplished correctly at all? Am I a fraud, now? I wasn’t before…

It’s a really painful trade for me.

Re: I am leaving the AI party after one drink

#102

Post by thepra »

That's true for new projects, I've found that as a bootstrap mechanism it is not ideal, takes away your hard earned preferences/know-how abs leaves you barely engaged on the technical path. On the other side I already have quite big codebases, some legacy and inherited, there exist vast amount of context to work on and figure out stuff/fixes/bugs/improvements and Claude Code excels 95% of the time in figuring the existing flows and functionality, if a vague or precise issue is known that's facing some files elaborations or user work flows it can do a good job in pointing out what are, with a high chance, the issues.

Re: I am leaving the AI party after one drink

#103

Post by babagan0ush »

itomato wrote: post #51

What did you first order at the bar? Did it burn? Did you become a teetotaler? Latecomers lack the hundreds of iterations and the experience that comes with it. The senses haven't been trained. There's a business here. Not one I want to be in, or one without major ethical drag, but a viable one. Fend off extinction, get fit.

[dead]

Re: I am leaving the AI party after one drink

#104

Post by legitster »

jplusequalt wrote: post #97
legitster wrote: post #93

Earlier quoted context omitted.

Here's a practical example: I am using AI to help me with my garden. It's been amazing - it helps me identify plants, identify soil issues, what fertilizer to use and what days to apply it, etc. What exactly did AI take from me? Spending hours of research on Google and Youtube to glean little incomplete bits and pieces? Calling a yard service? It's also clearly obvious when AI gives bad or incorrect advice - I am sti…

>What exactly did AI take from me? Spending hours of research on Google and Youtube to glean little incomplete bits and pieces? Calling a yard service? An opportunity for a deeper understanding of gardening? If you spend hours researching on gardening and come away with an incomplete understanding of what you were attempting to do, I'm not sure that's immediately the fault of the research available. It could be that…

> It could be that you just didn't do a good job searching for the necessary information.

How can you? The existing resources are terrible.

> But I think it would be a step to far to say that a smattering of specific questions can replace the sturdy foundation povided by a typical education--e.g. through apprenticeship, books, etc.

I am not going to go through a college program for my own garden. And I have books! But unless you can read a tiny and perform a small research project, you are not going to know how all of the plants in your specific garden in your specific region in your specific weather are going to behave.

The best I could do is hire an expert - but again I am learning less by hiring it out.

> Also, __how__ could you make such a statement unless you already possessed the knowledge ahead of time to make such a judgment?

"Use X to kill the moss". It didn't kill the moss. I will now use AI to find a list of alternative things to try to kill the moss, and learn what works in my garden.

The idea that AI is going to make people stop learning I don't think is born out in practice. It might make some people stop researching as an activity though.

Re: I am leaving the AI party after one drink

#105

Post by legitster »

MisterTea wrote: post #88
legitster wrote: post #7

> I don’t want to feel this kind of “addiction.” > I don’t want to depend on something doing the work I earn money with. > I don’t want to give up my brain and become lazy and not think for myself anymore. There are a lot of good reasons we should be skeptical of AI and not give up on essential skills. But sometimes I want to shake these people by the shoulders. Do you drive an automatic car? Do you use a microwave?…

> Do you drive an automatic car? Do you use a microwave? Do you buy food from a grocery store? Do you own power tools? None of these things allow you to turn your brain off while the machine does the work. I still have to DRIVE the car and all the thinking that goes with that. It's not a robotaxi. I still have to acquire and prep the food I am microwaving. It's not a replicator. I still have to know what I want to ea…

All of your examples involve you sitting and waiting. That doesn't seem like an apt analogy for what AI can do. You don't have to sit there and come up with other things to do while the AI does the work.

When AI works (and technology in general) that's kind of what it's like. You'll never perceive that you are not doing the work anymore because you won't perceive the work.

Re: I am leaving the AI party after one drink

#106

Post by ThrowawayR2 »

legitster wrote: post #59
tines wrote: post #16

Earlier quoted context omitted.

The problem is that you're likening fundamentally unlike things. AI isn't like a microwave or an automatic car or a power tool. It does not augment you. As I said elsewhere: AI is not a bicycle for the mind, it's an easy chair. You will lose more than you ever gain.

This is purely a matter of perception. Cooking a meal is a deeply intellectual process. If I buy a meal from a restaurant, yes I am losing a skill. But if making a hollandaise is not a skill I ever need in my life, it's not really a practical loss. AI is taking problems and putting them in a drawer so we never have to think about it again. Matches de-intellectualized making a fire. A washing machine de-intellectualiz…

> "It only a loss if you think the skill and ability you are losing is intrinsically valuable..."

I'm fascinated by the AI bros putting hollandaise sauce and making fires on the same level as creating production software. One hopes that it is because they create only very simple software, making the analogy less invalid than it would be for more complex software. If not, the implication is that loss of the reasoning and cognitive ability needed to build foundational software like libraries and frameworks is not important to them.

The only thing that separates homo sapiens from other species is the sapience. Diminishing or atrophying one's own cognitive abilities is the same as climbing down the evolutionary ladder.

Re: I am leaving the AI party after one drink

#107

Post by ValentineC »

legitster wrote: post #78
cush wrote: post #70

Earlier quoted context omitted.

> making a hollandaise is not a skill I ever need in my life I know you just wanted to poke at the analogy, but if you like hollandaise, it's one of the easiest and most rewarding sauces to make at home! Restaurant hollaindaise is usually terrible

Agreed. (Though it's not as easy as a béchamel, and yet I still see people buy jarred alfredo sauces. You can literally make an amazing alfredo sauce with pantry ingredients in less time than it takes to boil the noodles! Why would anyone buy an alfredo sauce!?) Although this more or less is my point. If people are willing to give up these incredibly high reward, low effort skills - how much more uphill is the battle…

> Why would anyone buy an alfredo sauce!?

Ignorance aside, jarred sauces are sorta shelf stable, and I have occasionally run out of butter and milk.

Re: I am leaving the AI party after one drink

#108

Post by legitster »

ThrowawayR2 wrote: post #106
legitster wrote: post #59

Earlier quoted context omitted.

This is purely a matter of perception. Cooking a meal is a deeply intellectual process. If I buy a meal from a restaurant, yes I am losing a skill. But if making a hollandaise is not a skill I ever need in my life, it's not really a practical loss. AI is taking problems and putting them in a drawer so we never have to think about it again. Matches de-intellectualized making a fire. A washing machine de-intellectualiz…

> " It only a loss if you think the skill and ability you are losing is intrinsically valuable... " I'm fascinated by the AI bros putting hollandaise sauce and making fires on the same level as creating production software. One hopes that it is because they create only very simple software, making the analogy less invalid than it would be for more complex software. If not, the implication is that loss of the reasonin…

I mean, doesn't the fact that people rely on these libraries and frameworks without thinking it itself prove the value of intentionally compartmentalizing off skills?

No one is arguing that everyone needs to build programs ground up from assembly. So what's the magic difference between using a framework and asking a computer to write out the for-loops for me?

Re: I am leaving the AI party after one drink

#109

Post by saltcured »

Nevermark wrote: post #62

Some of us have been waiting our whole lives for a comprehensive DWIM command. > DWIM is an embodiment of the idea that the user is interacting with an agent who attempts to interpret the user's request from contextual information. Since we want the user to feel that he is conversing with the system, he should not be stopped and forced to correct himself or give additional information in situations where the correcti…

IMO, many such people, unfortunately, are prone to forget to state the addendum NWIS to express their true hope: Do What I Mean, Not What I Said.

And this is kind of terrifying to me, in the context of an LLM that is working completely based on What You Said and any ability to Do What You Mean has to come from murky associations in the training data.

It is more than kind of terrifying when this is then extended to scenarios requiring novel analysis and problem solving, rather than just performing a repetitive, idiomatic task for the N+1th time.

Re: I am leaving the AI party after one drink

#110

Post by andai »

andai wrote: post #26

Reposting a comment (and one of the replies) since it's relevant: --- It occurred to me on my walk today that a program is not the only output of programming. The other, arguably far more important output, is the programmer. The mental model that you, the programmer, build by writing the program. And -- here's the million dollar question -- can we get away with removing our hands from the equation? You may know that…

Update: I fleshed this out into an article:

The Deliverable is You! Programming as Theory Building

https://nekolucifer.substack.com/p/the-deliverable-is-you-pr...

The open question is to what degree the actual typing is necessary. I see it as proof that the mental model is sound.

The thinking produces the mental model, the mental model produces the program. In the absence of that final step, what validates the model? (Complaints from users and colleagues? ;)

Another great thread to pull on: the overlap between "moving up levels of abstraction", i.e. AI assisted programming merely continues and accelerates a trend that has been there for the entire history of programming.

Post reply on HN