[echelon]

> Truthfully, I don't want to get advice from people who become addicted to AI, sorry.

If you sleep on this, these people are going to take your job.

I've been writing serious systems code for 15 years. Systems that handled billions of dollars of transaction volume a day and whose hourly outages cost billions of dollars. These are systems you have to design carefully. Active-active, beyond five nines reliable.

I'm telling you AI is extremely beneficial even in this segment of the market. The value prop is undeniable.

I'm easily getting twice my workload done with AI, and I'm not even leveraging the full extent of the tools. I've only just started to do more than fancy tab-autocomplete.

This is going to be a huge shift in our industry, and I would brace for impact.

[falloutx]

I dont want to take responsibility for 10 features at the same time because one guy on the internet is able to bullshit the execs.

> I'm easily getting twice my workload done with AI, and I'm not even leveraging the full extent of the tools. I've only just started to do more than fancy tab-autocomplete.

Are you getting 2x the money?

[jfyi]

> Are you getting 2x the money?

Probably no and additionally paying $200/mo out of pocket for the pleasure of doing twice the work.

[brailsafe]

Why would it be out of pocket?

[cube00]

Companies won't even pay for IntellJ, why would they pay for the top tier Claude plan so you can run the 10 agents you need? (and yes it needs to be multiple agents according to their founder [1])

[1]: https://news.ycombinator.com/item?id=46470017

[JohnLeitch]

>I'm easily getting twice my workload done with AI, and I'm not even leveraging the full extent of the tools.

It seems that every person who says this never elaborates on the nature of their work. What exactly are you writing? What languages? Technologies? What does the LLM assist with? In what ways does it hamper more than help?

I ask these questions because I have yet to see any meaningful, real world application of AI at my job. There's definitely interest, but every exploratory effort seems to fall flat, sometimes comedically so e.g. recently we had Sonnet 4.5 recommend some JavaScript for a UI hang we were looking into. It also recommended we use WebWorkers to improve perf. Sounds great. Looks great, with nice markdown and whatnot. Too bad it was a legacy MFC application written in C and C++.

[sph]

I don’t want to do twice my workload. I’m old enough to have learned that the faster and more efficient you are, the more demands they pile on you, and the net result is more stress, more expectations for the same pay. AI doesn’t solve unreasonable demands, shifting requirements and looming deadlines, does it?

And still, writing code is not even the bottleneck, the thinking, meeting stakeholders, figuring out technical problems is. What would I do with a machine that spits out bad code.

I guess I’m not cut out for a field where the only metric that counts is how many tickets and lines of code one can churn out in an hour any more.

[gtowey]

In my 20+ years in the tech industry, the most important thing I have learned is that writing code is the least significant part of being a software engineer.

You write code once, but then it's modified endlessly. Supporting and using and improving the code is the majority of the work. LLMs are terrible for that. They tend to write code that isnt easy to read and modify. They don't plan for future use cases. They often paint themselves into a corner with successive modifications, and lose context as the project grows.

Don't worry about your job just yet. We are a long long way from replacing people.

[realusername]

Personally I get a +5% productivity in a good day with AI.

I do double my productivity on personal projects but they aren't entreprise style jobs.

I really hope for those AI companies that my situation isn't too common because burning billions to make dev hobbies more productive doesn't sound too good of a business plan.

[atherton94027]

Curious, could you give examples of how you've been able to double your productivity with AI?

[Cthulhu_]

In my (not systems engineering) opinion, most time spent writing code is boilerplate and rituals; unit tests are pretty repetitive, creating a React component is a lot of repetition, etc. A LLM code assistant can do these boring things faster.

[atherton94027]

Yeah I agree with you on that, I'm just curious about the systems programming use case as in my experience you have to think deep about interactions and working with an agent blunts that

[wbsun]

These people are not going to take your job, the people who uses tools smartly while having the knowledge and experience in highly reliable distributed systems are. If human in the loop is not required any more, nobody is going to keep their job.

[bfung]

Totally agree.

The uncomfortable truth is that the skill of _writing_ code is becoming commoditized.

Reading, understanding, designing efficient systems, and planning changes based on that, at least for now, will still be for human experts (and AI already a great assist here).

But churn out yet another webpage/website? People doing this will need to move on from this as their primary job.

[falloutx]

>But churn out yet another webpage/website?

People exaggerate how many people work on just simple websites / webpages. Anyone needing a website has been able to get it for very cheap for a long time.

[whstl]

Yep. It has been the case forever. The reason we keep/kept getting new tools is because laymen would exhaust their capabilities in the old tools and started requiring a developer once they need THAT ONE feature.