Effects of Gen AI on High Skilled Work: Experiments with Software Developers • Hacker News Board

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#431

Post by highfrequency » Sat, Sep 07, 2024, 4:24 PM UTC

1dom wrote: post #326

IgorPartola wrote: post #254

Earlier quoted context omitted.

I think there is another bit to this which is cargo cult tendencies. Basically DevOps is a great idea under certain circumstances and works well for specific people in that role. Realistically if you take a top talent engineer they can likely step into any of the three roles or even some others and be successful. Having the synergy of one person being responsible for the boundary between two roles then makes your ROI…

> There is no methodology (organizational, programming, testing, etc.) that can make excellence out of mediocrity. This is a thought provoking phrase, and I liked that. Thinking a bit deeper, I'm not sure if it's accurate, practical or healthy though. I've seen mediocrity produce excellent solutions, and excellence churn out rubbish, and I suspect most people with a few years and tech jobs under their belt have. You…

Your emphasis on these two ideas resonates with me:

1. Optimal practices are highly context specific.

2. The search space is very high dimensional and more often than not we are blind to most of it.

Care to elaborate on these points with anecdotes?

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#432

Post by anotherburner23 » Sat, Sep 07, 2024, 4:32 PM UTC

jillesvangurp wrote: post #348

29athrowaway wrote: post #61

The study doesn't cover the long term effects of using generative AI on a project, which is the deskilling of your workforce. Because development will become an auction-like activity where the one that accepts more suggestions wins.

I look at this very differently. There are a lot of grumpy people jealously guarding their "skills" and artisanal coding prowess getting really annoyed at the juniors that come in and devalue what they do by just asking an AI to do the same thing and then moving on with their day. Young people are more mentally agile and most young people are now growing up with LLMs as a tool chain that was always there. I'm actuall…

"Tell me how you use tools to your advantage; not how you have a tool aversion or are unwilling to learn new things" is a false binary. It leaves out the possibility that the tool in question is not suited for its purpose.

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#433

Post by huijzer » Sat, Sep 07, 2024, 4:42 PM UTC

dimal wrote: post #402

It matters what you measure. The studies only looked at Copilot usage. I’m an experienced engineer. Copilot is worse than useless for me. I spend most of my time understanding the problem space, understanding the constraints and affordances of the environment I’m in and thinking about the code I’m going to write app. When I start typing code, I know what I’m going to write, and so a “helpful” Copilot autocomplete is…

> So, contrary to this report, I think that if experienced developers use AI well, they could benefit MORE than inexperienced developers.

A psychology professor I know says this holds in general. For any new tool, who will be able to get the most benefits out of it? Someone with a lot of skill already or someone with fewer skill? With less skill, there is even a chance that the tool has a negative effect.

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#434

Post by sdiupIGPWEfh » Sat, Sep 07, 2024, 4:44 PM UTC

neilv wrote: post #78

choilive wrote: post #60

Earlier quoted context omitted.

Its been a while since my undergrad (>10 yrs) but many of my peers were majoring in CS or EE/CE because of the money, at the time I thought it was a bit depressing as well. With a few more years under my belt I realized theres nothing wrong with doing good work and providing yourself/your family a decent living. Not everyone needs the passion in their field to become among the best or become a "10x"er to contribute.…

Yeah, I think that the quality of work (skill + conscientiousness), and the motivation for doing the work, are two separate things. Off-the-cuff, three groups: 1. There are people who are motivated by having a solid income, yet they take the professionalism seriously, and do skilled rock-solid work, 9-5. I'd be happy to work with these people. 2. There are other people who are motivated by having a solid or more-than…

Blend of #1 & #3 here. Without the pay, I don't think I'd muster the patience for dealing with all the non-programming BS that comes with the job. So I'd have found something else with decent pay, and preferably not an office job. Sometimes I wish I'd taken that other path. I have responsibilities outside work, so I'm rarely putting in more than 40 hours. Prior to family commitments, I had side projects and did enough coding outside work. I still miss working on hobby video game projects. But I do take the professionalism seriously and will do the dirty work that has to be done, even if means cozying up to group #2 to make things happen for the sake of the project.

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#435

Post by heed » Sat, Sep 07, 2024, 4:59 PM UTC

dimal wrote: post #402

It matters what you measure. The studies only looked at Copilot usage. I’m an experienced engineer. Copilot is worse than useless for me. I spend most of my time understanding the problem space, understanding the constraints and affordances of the environment I’m in and thinking about the code I’m going to write app. When I start typing code, I know what I’m going to write, and so a “helpful” Copilot autocomplete is…

I like to use copilot when writing tests. It's not always perfect but makes things less tedious for me.

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#436

Post by godelski » Sat, Sep 07, 2024, 5:05 PM UTC

NortySpock wrote: post #229

godelski wrote: post #123

Earlier quoted context omitted.

- my laundry app (that I must use) takes minutes to load, doesn't cache my last used laundry room, and the list of rooms isn't even fucking sorted (literally: room 3, room 7, room 1, room 2, ...) - my AC's app takes 45 s to load even if I just used it, because it needs to connect. Worse, I'll bring the temp down in my house and in the evening raise it, but it'll come on even when 5F below my target value, staying on…

> my laundry app (that I must use) takes minutes to load Do they have a place to mail your complaints? Who forces you to use the app? Annoy them. Annoy their boss. Write a flier pointing out how long it takes to use the app and leave those fliers in every laundry room you visit (Someone checks on those machines, right?). Heck, send an organization-wide email pointing out this problem. (CC your mayor, council-member,…

Short of of writing to my mayor, I've done most of that. Big motivation to learn reverse engineering and some basic hacking.

I can't hack everything.

I can't fix everything.

I need time for my own things I need time to be human.

But no one is having it. My impression is that while everyone else is annoyed by things there's a large amount of apathy and acceptance for everything being crud.

I know you're trying to help, but this is quite the burden on a person to try to fix everything. And things seem to be getting worse. So I'm appealing to the community of people who create the systems. I do not think these are things that can be fixed by laws. You can't regulate that people do their best or think things through.

I appeal here because if I can convince other developers that slowing down will speed things up, then many more will benefit (including the developers and the companies they work for). Even convincing a few is impactful. Even more when the message spreads.

Small investments compound and become mighy, but so do "shortcuts"

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#437

Post by eric_h » Sat, Sep 07, 2024, 5:17 PM UTC

heed wrote: post #435

dimal wrote: post #402

It matters what you measure. The studies only looked at Copilot usage. I’m an experienced engineer. Copilot is worse than useless for me. I spend most of my time understanding the problem space, understanding the constraints and affordances of the environment I’m in and thinking about the code I’m going to write app. When I start typing code, I know what I’m going to write, and so a “helpful” Copilot autocomplete is…

I like to use copilot when writing tests. It's not always perfect but makes things less tedious for me.

I recently switched to cursor, and am in the process of wrangling an inherited codebase that had effectively no tests and cursor has saved me _hours_ it's generally terrible at any actual code refactoring, but it has saved me a great deal of typing while adding the missing test coverage.

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#438

Post by sroussey » Sat, Sep 07, 2024, 5:42 PM UTC

about3fitty wrote: post #403

bluepizza wrote: post #371

Earlier quoted context omitted.

It is not a switch up. The author said that an engineer would find these during their journey, and I am asking when. I don't have a strongly held opinion here, I literally want to know when. I am curious about it.

I found them in 2015 when I was maintaining a legacy app for a university. The developer that implemented them could have used a few bools but decided to cram it all into one byte using bitwise operators because they were trying to seem smart/clever. This was a web app, not a microcontroller or some other tightly constrained environment. One should not have to worry about solar flares! Heh.

> trying to seem smart/clever.

Maybe. Every time I write something I consider clever, I often regret it later.

But young people in particular tend to write code using things they don’t need because they want experience in those things. (Just look at people using kubernetes long before there is any need as an example). Where and when this is done can be good or bad, it depends.

Even in a web app, you might want to pack bools into bytes depending on what you are doing. For example, I’ve done stuff with deck.gl and moving massive amounts of data between webworkers and the size of data is material.

It did take a beat to consider an example though, so I do appreciate your point.

Coming from a double major including EE though, all I have to say is that everyone’s code everywhere is just a bunch of NOR gates. Now if you want to increase your salary, looking up why I say “everything is NOR” won’t be useful. But if you are curious, it is interesting to see how one Boolean operand can implement everything.

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#439

Post by brandall10 » Sat, Sep 07, 2024, 5:42 PM UTC

dimal wrote: post #402

It matters what you measure. The studies only looked at Copilot usage. I’m an experienced engineer. Copilot is worse than useless for me. I spend most of my time understanding the problem space, understanding the constraints and affordances of the environment I’m in and thinking about the code I’m going to write app. When I start typing code, I know what I’m going to write, and so a “helpful” Copilot autocomplete is…

Copilot isn't particular useful. At best it comes up with small snippets that may or may not be correct, and rarely can I get larger chunks of code that would be working out of the gate.

But Claude Sonnet 3.5 w/ Cursor or Continue.dev is a dramatic improvement. When you have discrete control over the context (ie. being able to select 6-7 files to inject), and with the superior ability of Claude, it is an absolute game changer.

Easy 2-5x speedup depending on what you're doing. In an hour you can craft a production ready 100 loc solution, with a full complement of tests, to something that might otherwise take a half day.

I say this as someone with 26 yoe, having worked in principal/staff/lead roles since 2012. I wouldn't expect nearly the same boost coming at less than senior exp. though, as you have to be quite detailed at what you actually want, and often take the initial solution - which is usually working code - and refine it a half dozen times into something that you feel is ideal and well factored.

Re: Effects of Gen AI on High Skilled Work: Experiments with Software Developers

#440

Post by kbaker » Sat, Sep 07, 2024, 5:54 PM UTC

dimal wrote: post #402

It matters what you measure. The studies only looked at Copilot usage. I’m an experienced engineer. Copilot is worse than useless for me. I spend most of my time understanding the problem space, understanding the constraints and affordances of the environment I’m in and thinking about the code I’m going to write app. When I start typing code, I know what I’m going to write, and so a “helpful” Copilot autocomplete is…

Yeah, Copilot is meh. Aider-chat for things with GPT-4 earlier this year was a huge step up.

But recently using Claude Sonnet + Haiku through OpenRouter also with aider, and it is like a new dimension of programming.

Working on new projects in Rust and a separate SPA frontend, it just ... implements whatever you ask like magic. Gets it about 90-95% right at the first prompt. Since I am pretty new to Rust, there are a lot of idiomatic things to learn, and lots of std convenience functions I don't yet know about, but the AI does. Figuring out the best prompt and context for it to be effective is now the biggest task.

It will be crazy to see where things go over the next few years... do all junior programmers just disappear? Do all programmers become prompt engineers first?