[pojzon]

AI changes the landscape alot.

Even as an experienced engineer the sheer amount of knowledge you get access to from a single prompt without having to read any tutorial or book can change a Junior into Senior very quickly.

I just hope it wont face the same issue google had:

- starting with good answers and evolving into mud

[pjerem]

For now, my feelings are that AI can be really helpful if you know the field. You need to be senior to know how to ask things and to understand the answers.

For anyone not understanding the domain, AI is just a glorified stack overflow but with hallucinations.

Hallucinations can be fought but you need to suggest that you got the wrong answer and why. And that requires a deep understanding of the domain.

Honestly I’m really frightened that we collectively accept that AI is an acceptable source of truth and that it’s ok to make decisions from its output or worse, use it as a learning material.

[axus]

Claude, compose an email response to shift the blame from my latest AI-induced screwup.

[pojzon]

Point was - it will speed up junior to senior transition. Not that it will replace seniors with juniors.

We will also need less senior ppl, coz AI will be "good enough", most stuff we are faced with isn't really that complex.

[timr]

You don't get anywhere faster "learning" from something that lies to you 20% of the time.

It's a bit like working with a bad colleague who is very fast, but very arrogant. You can't trust what they say because they're wrong often enough to make costly mistakes common. But you can't fight them on every little thing, either. The only solution is to already be an expert, and ignore them when they're wrong.

I honestly believe AI -- if it has a dramatic impact at all -- will only reduce the value of junior employees.

[mulmen]

My entire formal education and subsequent career stands in opposition to this statement. Unless you mean that learning requires being lied to more than 20% of the time.

[timr]

That sounds like exaggeration to me, in service of a bias against "formal education". But OK. YMMV.

My statement applies only to the experience of working with the things, and relates not at all to "formal education". If I have to learn the subject to debug what they're putting out, then the rate-limiting step of using them is...learning the subject. Same as it ever was.

Having a stochastic parrot spit a stream of 20% nonsense at me doesn't make learning go faster -- it definitely does make work go faster if I'm already an expert, however.

[mulmen]

I actually think the time I spent in school was valuable and that formal education gets a bad rap around here. I believe that my teachers all had good intentions. But they weren't always right. In my experience nobody is right 80% of the time.

I'm skeptical of the AI hype but I do believe there is value. Similar to self driving cars an AI assistant or teacher doesn't have to be right all the time, it only has to be right more often. Proper use of this tool will require skillbuilding like anything else.

[mulmen]

If AI is “good enough” but you need to be senior to use it don’t we need less juniors? Like, just enough to replenish the retiring seniors.

[pjerem]

That’s my point, yes.

At least in the current state of affairs.

Also it’s a technical view from someone who played/worked with AI for more than a year.

It may not be the opinion of upper management :) And upper management can survive a pretty long time after their own mess.

[greedo]

AI won't be as impactful in generating good results (though it may result in middle management firing devs/admins since they think they're easily replaceable).

As an example, I asked claude.ai to generate an ansible playbook for patching a Docker cluster. It created one, something I'd expect a junior admin to be able to whip up pretty easily. Then I noticed something funky. The playbook was OS agnostic, just using some where clauses to handle debian, RHEL, etc. Nice stuff, but wrong. One of the tasks was to clean the cache after apply updates. The AI got apt correct (autoclean), but assumed incorrectly that yum had the same parameter. Just a small detail that would have shown up when the playbook was run, but when it can't handle the small details, why bother?

[NortySpock]

Because it wrote 90% of the code and you had two quick things to inspect and fix?

Sounds like a healthy tradeoff to me.

[greedo]

It wrote a template that passed a playbook linter. Whether any of the tasks were actually correct is highly questionable. I caught the errors because I have lots of experience. Maybe we need an AI to double check the work of the first AI?

/s

[mrcode007]

It only works for established technologies. New inventions are by definition not in the training dataset. What then?

[bdcravens]

Unfortunately the majority of jobs aren’t really innovation roles.

[smrtinsert]

> can change a Junior into Senior very quickly

Not to derail the thread, but its really not that simple. People are subject to their own biases and knowledge gaps. Long form responses do not fix either of those issues though they might help with the latter. If it really upskilled everyone that cleanly its effect on roles would've happened by now. I've noticed a lot of people frankly do not have the natural "context length" needed to really take advantage of AI and there are those that are really benefitting from it, but they're in the smallest minority.

[pksebben]

Man, comment was greyed between reading and clicking reply - I get that the back-and-forth over AI is a thing but if you're gonna respond negatively to an arguable point based on a topic with this much nuance, at least respond with words as well as downvotes. ANYWAY...

...It's that very google issue that convinces me the most important thing we can do is build towards locally runnable LLMs. There's no reason we can't eventually get to gpt4-or-better on consumer hardware, it's just going to take some time and research and we're gonna have to dodge the incessant ladder-pulls that the big dogs keep lobbying for ("AI is dangerous! Only we can be trusted(tm)!").

You can't enshittify an application I own, that's running on my hardware. Not with the current state of things, at least.