I use LLMs in coding. There are Junior Devs in my team.

If you think LLMs operate at "junior dev" capacity you either don't work with junior devs and is just bullshitting your way around here, or you just pick pretty awful junior devs.

LLMs are alright. An okay productivity tool, although its inconsistencies many times nullify productivity gains - By design they often spit out wrong results that look and sound very plausible. A productivity blackhole. Its mistakes are sometimes hard to spot, but pervasive.

Beyond that, if your think that all a dev does is spit out code, and since LLMs can spit out code it can replace devs in some imaginary timeline, you are sorely mistaken. The least part of my work is actually spitting out code, although it is the part I enjoy the most.

I honestly feel way nore threatened by economic downturns and the looming threat of recession. The only way LLMs threaten me is by being a wasteful technology that may precipitate a downturn in tech companies, causing more layoffs, etc nd so forth.

If you define "junior" based mostly on age, then LLM's aren't yet at the level of a good "junior".

If you base it on ability, then an LLM can be be more useful to a good developer than 1 or more less competent "junior" team members (regardless of their age).

Not because it can do all the things like any "junior" can (like make coffee), but because the things it can do on top of what a "junior" can do, more than makes up for it.

The value of developers is not the code they output. It's the mental models they develop of the problem domain and the systems they build. LLMs can output code without developing the mental models.

Code is liability. The knowledge inside developers' heads is the corresponding asset. If you just produce code without the mental models being developed and refined, you're just increasing liability without the counterpart increase in assets.

>> If you think LLMs operate at "junior dev" capacity you either don't work with junior devs and is just bullshitting your way around here, or you just pick pretty awful junior devs.

I’ve hired lots of junior devs, some of them very capable. I’ve been in this industry for more than 15 years. LLMs operate at junior dev capacity, that’s pretty clear to me at this moment.

I sincerely doubt both your experience and your ability to hire decent devs.

I know, it's an highly unpopular opinion among devs. Let's revisit this comment in 5 years...

I sincerely doubt your ability to use LLMs well.

Yep. There are people who love programming, it's the best part of the work anyhow! And then there are people who come and tell that whatever you do doesn't matter and they are more content on getting a new app by writing a prompt and deploying possibly buggy code. Two different crowds of people.

I'm in a middle. I enjoy Zed and its predictions, I utilize R1 to help me to reason. I do _not_ ever want to stop programming. And I see so often whenever somebody less experienced than me shows me look how Cursor did this with three prompts, can we merge? And the solution is just wrong and doesn't solve the hard issues.

For me the biggest issues are the people who want to see the craft of programming gone. But I do enjoy the tooling.

Nah. "AI" is just really, really lame and square. People have a visceral reaction to it even when it's actually not that bad.

These types of articles are just catching the next meme wave, which will be hating on and making fun of "AI" of all sorts.

> it seems likely that LLMs will obsolete precisely the skills that developers use to earn their income

I’m not particularly worried. I think it’s obvious that software engineering is definitely an “intelligence complete” problem. Any system that can do software engineering can solve any problem that requires intelligence. So, either my job is safe or I get to live through the fall of almost all white collar disciplines. There’s not a huge middle ground.

Although perhaps this is just the programmer stereotype of thinking that if someone can code, they can do anything.

> Any system that can do software engineering can solve any problem that requires intelligence. So, either my job is safe or I get to live through the fall of almost all white collar disciplines. There's not a huge middle ground.

How about the middle ground where a human using AI replaces you?

The human job is (maybe) safe, but your job?

Of course that is exactly the middle ground that I’m not certain is so big.

Developer productivity has gone up immensely in the last 50 years and the industry is larger than ever.

Um. How are you measuring that productivity?

Any meaningful metric.