[reedf1]

I think it is possible that the widespread introduction of ChatGPT will cause a brief hiatus on hiring due to the inelasticity of demand. For the sake of argument, imagine that ChatGPT makes your average developer 4x more productive. It will take a while before the expectation becomes that 4x more work is delivered. That 4x more work is scheduled in sprints. That 4x more features are developed. That 4x more projects are sold to clients/users. When the demand eventually catches up (if it exists), the hiring will begin again.

[TSiege]

I am not asking this as a gotcha, but a genuine curiosity for you or other people who find AI is helping them in terms of multiples. What is your workflow like? Where do you lean on AI vs not? Is it agentic stuff is tab by cursor?

I find AI helpful but no where near a multiplier in my day to day development experience. Converting a csv to json or vis-versa great, but AI writing code for me has been less helpful. Beyond boiler plate, it introduces subtle bugs that are a pain in the ass to deal with. For complicated things, it struggles and does too much and because I didn't write it I don't know where the bad spots are. And AI code review often gets hung up on nits and misses real mistakes.

So what are you doing and what are the resources you'd recommend?

[SatvikBeri]

I get very good results from Claude Code, something like a 3x. It's enough that my cofounders noticed and commented on it, and has had a lot of measurable results in terms of saving $ on infrastructure.

The first thing I'll note is that Claude Code with Claude 4 has been vastly better than everything else for me. Before that it was more like a 5-10% increase in productivity.

My workflow with Claude Code is very plain. I give it a relatively short prompt and ask it to create a plan. I iterate on the plan several times. I ask it to give me a more detailed plan. I iterate on that several times, then have Claude write it down and /clear to reset context.

Then, I'll usually do one or more "prototype" runs where I implement a solution with relatively little attention to code quality, to iron out any remaining uncertainties. Then I throw away that code, start a new branch, and implement it again while babysitting closely to make sure the code is good.

The major difference here is that I'm able to test out 5-10 designs in the time I would normally try 1 or 2. So I end up exploring a lot more, and committing better solutions.

[reedf1]

4x is a number I pulled out of thin air. I'm not sure I even yet believe there is a net positive effect of using AI on productivity. What I am sure about in my own workflow is that is saves me time writing boilerplate code - it is good at this for me. So I would say it has saved me time in the short-term. Now does not writing this boilerplate slow me down long-term? It's possible, I could forget how to do this myself, some part of my brain could atrophy (as the MIT study suggests). How it affects large teams, systems and the transfer of knowledge is also not clear.

[dgfitz]

I read this sentiment a lot, and it is true for me too as a completely average software engineer.

Makes it seem like the actual problem to be solved is reducing the amount of boilerplate code that needs to be written, not using an LLM to do it.

I'm not smart enough to write a language or modify one, so this opinion is strongly spoken, weakly held.

[eru]

I wouldn't be too worried about the atrophy. Or at least not much more than you already were: you get the same atrophy effect just from IDEs and compiler errors and warnings.

To give a concrete example: I'm pretty good at doing Python coding on a whiteboard, because that's what I practiced for job interviews, and when I first learned Python I used Vim without setting up any Python integration.

I'm pretty terrible at doing Rust on a whiteboard, because I only learned it when I had a decent IDE and later even AI support.

Nevertheless, I don't think I'm a better programmer in Python.

[fcatalan]

I use it a lot for reducing friction. When I procrastinate about starting something I ask the AI to come up with a quick plan. Maybe I'll just follow the first step, but it gets me going.

Sometimes I´ll even go a bit crazy on this planning thing and do things a bit similar to what this guy shows: https://www.youtube.com/watch?v=XY4sFxLmMvw I tend to steer the process more myself, but typing whatever vague ideas are in my mind and ending up in minutes with a milestone and ticket list is very enabling, even if it isn´t perfect.

I also do more "drive by" small improvements:

- Annoying things that weren't important enough for a side quest writing a shell script, now have a shell script or an ansible playbook.

- That ugly CSS in an internal tool untouched for 5 years? fixed in 1 minute.

- The small prototype put into production with 0 documentation years ago? I ask an agentic tool to provide a basic readme and then edit it a bit so it doesn´t lie, well worth 15 minutes.

I also give it a first shot at finding the cause of bugs/problems. Most of the time it doesn't work, but in the last week it found right away the cause of some long standing subtle problems we had in a couple places.

I have also had sometimes luck providing it with single functions or modules that work but need some improvement (make this more DRY, improve error handling, log this or that...) Here I´m very conservative with the results because as you said it can be dangerous.

So am I more productive? I guess so, I don't think 4x or even 2x, I don't think projects are getting done much faster overall, but stuff that wouldn't have been done otherwise is being done.

What usually falls flat is trying to go on a more "vibe-coding" route. I have tried to come up with a couple small internal tools and things like that, and after promising starts, the agents just can't deal with the complexity without needing so much help that I'd just go faster by myself.

[alyandon]

I lean a bit on LLMs now for initial research/prototype work and it is quite a productivity boost vs random searches on the web. I generally do not commit the code they generate because they tend to miss subtle corner cases unless the prompts I give them are extremely detailed which is not super useful to me. If an LLM does produce something of sufficient quality to get committed I clearly mark it as (at least partially) LLM generated and fully reviewed by myself before I mash the commit button and put my name on it.

Basically, I treat LLMs like a fairly competent unpaid intern and extend about the same level of trust to the output they produce.

[ninetyninenine]

Don’t ask the agent to do something complex. Break it down into 10 manageable steps. You are the tester and verifier of each step.

What you will find is that the agent is much more successful in this regard.

The LLM has certain intrinsic abilities that match us and like us it cannot actually code 10,000 lines of code and have everything working in one go. It does better when you develop incrementally and verify each increment. The smaller the increments the better it performs.

Unfortunately the chain of thought process doesn’t really do this. It can come up with steps, sometimes the steps are too big and it almost never properly verifies things are working after each increment. That’s why you have to put yourself in the loop here.

Like allowing the computer to run test and verify an application works as expected on each step and to even come up with what verification means is a bit of what’s missing here and I think although this part isn’t automated yet, it can easily be automated where humans become less and less involved and distance themselves into a more and more supervisory role.

[alyandon]

Spot on - that is exactly my experience when working with LLMs.

[ianm218]

I'm in the same boat of some of the other commenters using Claude Code but I have found it atleast a 2X in routine backend API development. Most updates to our existing APIs would be on the order of "add one more partner integration following the same interface here and add tests with the new response data". So it is pretty easy to give it to claude code, tell them where to put the new code, tell it how to test, and let it iterate on the tests. So something that may have taken a full afternoon or more to get done gets done much faster and often with a lot more test coverage.

[ulrikrasmussen]

I have the same experience as you. It has definitely increased the speed with which I can look up solutions to isolated problems, but for writing code using agents and coming up with designs, the speed is limited by the speed with which I as a human can perform code reviews. If I was surrounded by human 10x developers who wrote all the code for me and left it for me to review it, I doubt my output would be 4x.

[TheDong]

I've personally managed to produce roughly 8x the production outages and show-stopper bugs than I did before LLMs, so thing are looking pretty good!

[Aperocky]

The competitive edge is now knowing how to debug all of those issues. Unfortunately not usually a skill possessed for entry level.

[postalrat]

So 4x more productive from WFH and 8x more from LLMs. The standard is now 32x more productive than 5 years ago.

[bluefirebrand]

And yet the sprint estimates haven't budged

Weird

[pseufaux]

And yet, you can still claim 100% of your code to be bug free :D

[ai-christianson]

We just shipped a major feature on our SaaS product. We, of course, used AI extensively.

The thing is, this feature leaned on every bit of experience and wisdom we had as a team --things like making sure the model is right, making sure the system makes sense overall and all the pieces fit together properly.

I don't know that "4x" is how it works --in this case, the AI let us really tap into the experience and skill we already had. It made us faster, but if we were missing the experience and wisdom part, we'd just be more prolific at creating messes.

[MajimasEyepatch]

But presumably you could have built it before, just slower, which is the point. For now, that speed-up just looks like a win because it’s novel, but eventually the speed-up will be baked into people’s expectations.

[ai-christianson]

Right, I'm just pointing out that if you're "never going to get there anyway," then going 4x faster isn't going to help.

[lazide]

The biggest issue at most companies is making a lot of heat and noise while not actually delivering effectively, so I expect AI to make this problem worse.

Usually, companies benefit more from slowing down and prioritizing, not ‘going faster’.

[eru]

> For now, that speed-up just looks like a win because it’s novel, but eventually the speed-up will be baked into people’s expectations.

It will still be a win: the rewards for the new productivity have to go somewhere in the economy.

Just like other productivity improvements in the past, it will likely be shared amongst various stakeholders depending on a variety of factors. The workers will get the lion's share.

[bluefirebrand]

> rewards for the new productivity have to go somewhere in the economy

Unless it goes to me I'm not entirely sure why I should care

I'll keep doing things the old way thanks, unless I personally get some benefit from it

"You are more productive but compensated the same" just shows how many of you are suckers

[eru]

Compare and contrast https://pseudoerasmus.com/2017/10/02/ijd/

[layer8]

> The workers will get the lion's share.

This doesn’t seem to be supported by history.

[eru]

Labour share have GDP has held roughly constant throughout the ages, even as we saw massive productivity increases since the dawn of the industrial revolution.

[layer8]

How does that equate to the workers receiving the lion's share of the massive productivity increases?

Statistics like https://media.equality-trust.out.re/uploads/2024/07/incomedi... suggest that since 1970 the top 10% have profited more in their income than the bottom 50%.

[j1elo]

Things should get even out with the 4x salary increases we'll also get thanks to that extra productivity, right?

[bluefirebrand]

No, all there is in the future is 4x as many layoffs

[vevoe]

That makes sense to me. There's another post on the front page right now talking about shortening the work week (I haven't read it yet tbf, so I could be wrong about it's content) because of AI. People have been talking about shorter work weeks for a long time now, it just doesn't happen. What does happen is we get more done and the GDP goes even higher.

[landl0rd]

“Shortening the workweek” sounds pretty bad… some people will suggest literally anything before higher wages.

[bachmeier]

I think that in countries with longer workweeks, that would be an incredible thing. There was recently a story about Denmark raising the retirement age to 70. If you graduate college at 22 and work the average number of hours until age 70 in Denmark, that's the same as working until 59 in the US and 52 in Mexico. Shorter workweeks will almost certainly translate into longer careers.

(https://www.oecd.org/en/data/indicators/hours-worked.html Mexico 2226 hours/week, US 1804, Denmark 1394)

[eru]

Real wages are going up all the time.

[eru]

You can already take a job with a shorter work week or move a region or country where shorter work weeks are common.

[elmean]

omg 4x scrum master inbound we are gonna be so agile

[ai-christianson]

Moving very fast, but going nowhere.

[bborud]

It would be interesting to see some research on exactly how much sustained productivity boost programmers can get by using LLMs. The reason this is a bit complicated is that the code would have to pass certain quality metrics. In particular when it comes to structural soundness -- whether a piece of code is something you can build on and evolve, or if it is disposable.

I think different generations of programmers have different opinions on what is quality output. Which makes judging the quality of code very context dependent.

If I were to guess I probably get somewhere in the range 10% to 20% productivity boost from LLMs. I think those are pretty astonishing numbers. The last time I got this kind of boost was when we got web search engines and sites like stack exchange.

I would suspect that if people experience 100% or more productivity boost from LLMs, something is off. Either we have very different ideas about quality, or we are talking about people who were not very productive to begin with.

I also think that LLMs are probably more useful if you are already a senior developer. You will have a better idea of what to ask for. You will also be in a better position to guide de LLM towards good answers.

...which kind of hints at my biggest worry: I think the gen-z programmers are facing a tough future. They'll have a harder time finding jobs with good mentors. They're faced with unrealistic expectations in terms of productivity. And they have to deal with the unrealistic expectations from "muggles" who understand neither AI nor programming. They will lack the knowledge to get the most from LLMs while having to deal with the expectation that they perform at senior levels.

We already see this in the job market. There has been a slight contraction and there are still a significant portion of senior developers available. Of course employers will prefer more experienced developers. And if younger developers believe in the hype that they can just vibe-code their way to success, this is just going to get worse.

[am17an]

With all the tools around, I think I've maybe become 20% more productive, but 50% less happy in arguing and babysitting the LLMs.

[bluefirebrand]

This is a net negative. Especially if you aren't 20% more paid at the same time

Sounds like AI has landed you on burnout treadmill

[Workaccount2]

>That 4x more features are developed. That 4x more projects are sold to clients/users.

The absolute best outcome of LLMs, and frankly where it seems to be headed, is the death of bloated one-stop-shop-for-everyone software. Instead people will be able to use their computers more directly than ever, without having to use/figure out complicated unintuitive swiss army knife software to solve their problems.

LLMs today can already make people the exact tools they need with no extra feature bloat or useless expansive packages. An print shop who just resizes their photos and does some minor adjustments not available from free tier software is no longer a slave to paying adobe $40/mo to use <1% of Photoshop's capabilities. They now can have their own tailor made in-house program for free.

LLM's will not be slotted in to replace devs on adobes dev teams. They can't work on a photoshop size codebase. However they will likely cut demand for Photoshop. Very few people will mourn the death of having to pay monthly for software just because there is a language barrier between them and their computer.