Comment by ath3nd | Hacker News Reader

Comment by ath3nd | Hacker News Reader

Parent Full thread

You are not alone. There are plenty of us, see here:

- Claude Code is a Slot Machine https://news.ycombinator.com/item?id=44702046

- GPTs and Feeling Left Behind: https://news.ycombinator.com/item?id=44851214

- I also used the Emperor/clothes metaphor: https://news.ycombinator.com/item?id=44854649

And just so we are clear, in the only current actual study measuring productivity of experienced developers so far, it actually led to 19% decline in productivity. https://metr.org/blog/2025-07-10-early-2025-ai-experienced-o...

So, if the study showed experienced developers had a decline in productivity, and some developers claim gains in theirs, there is high chance that the people reporting the gains are...less experienced developers.

See, some claim that we are not using LLMs right (skills issue on our part) and that's why we are not getting the gains they do, but maybe it's the other way around: they are getting gains from LLMs because they are not experienced developers (skills issue on their part).

I'll wait for more studies about productivity, one data point is not solid foundation, there are a lot of people who want this to be true, and the models and agent systems are still getting better

I'm an experience (20y) developer and these tools have saved me many hours on a regular basis, easily covering the monthly costs many times over

There’s not enough data…which somehow proves your pre-existing bias which goes against the only data we have.

I'm offering my anecdotal evidence as contrary to the loss of productivity piece that has been making the rounds, not as justification that they are wrong, but that we don't have enough data

It's also not the only data, here's one with results in the other direction

https://medium.com/@sahin.samia/can-ai-really-boost-develope...

> Productivity was tracked through metrics such as completed tasks (pull requests), code commits, and successful builds.

Making untested garbage faster to check off tasks quicker. Reopen rate please? New bug task rate? Nobody looked...

> The study also monitored code quality via build success rates. Importantly, increased productivity did not come at the cost of more errors, showing that Copilot helped developers code faster and more accurately.

It builds therefore it works. And the test suite was also AI generated?

Feels like we're back in primary school learning programming.

Classic case of gaming the metrics.

You sound like you haven't given Copilot and friends a thorough chance or evaluation. If you had, you'd know that...

- people don't outsource, they pair-program, because these things cannot do complex tasks on their own

- they are quite good at running tests with coverage, inspecting the results, and fixing both the tests and code

- people make mistakes, expecting AI to be perfect is unreasonable, they are tools, not replacements

> Making untested garbage faster to check off tasks quicker

> Feels like we're back in primary school learning programming.

> Classic case of gaming the metrics.

plain view bias causes others to discount your opinion

> people make mistakes, expecting AI to be perfect is unreasonable, they are tools, not replacements

This is the key. These tools are an improvement for many people, but others pooh-pooh them for not being perfect. Working in a team with other programmers (or looking at my own older code) I often see mistakes, often obvious to me now.

Glad to see more data!

It’s interesting they showed the biggest gains for junior developers. The other study showing productivity losses for experienced developers. That suggests these tools are a lot more helpful for junior developers compared to senior developers at the moment.

From my experience, a senior dev who knows how to work these tools will be better than both. We need data on how long, and how much effort into developing the skill, working with AI a developer has put in.

Theore you use it, the more you get a feel for when and when not to use it

Your comments are citing this blog post and arxiv preprint.

You are also misrepresenting the literature. There are many papers about LLMs and productivity. You can find them on Google Scholar and elsewhere.

The evidence is clear that LLMs make people more productive. Your one cherry picked preprint will get included in future review papers if it gets published.

Good thing you linked to these articles that support your claims, otherwise I would be left feeling like you are misrepresenting the literature to push your own agenda.

You have the ability to search Google Scholar, I'm quite sure of it

I found nothing. Could you share the links you found, or are you lying

I do, and did, and found you were spreading misinformation.

This is false, but the nice thing is that anyone can go check it and don't have to rely on the claims of an account created specifically to reply to me.

This is false, this account was made specifically to inform people that a correct proof of P=NP is on Google Scholar.

Go check it!

> So, if the study showed experienced developers had a decline in productivity,

You forgot to add: first time users, and within their comfort zone. Because it would be completely different result if they were experienced with AI or outside of their usual domain.

Reply on news.ycombinator.com