will AI wipe out IT jobs

[PolyPeter]

9 hours ago, bazera said:

I'm not sure about the begginers, it's really hard to get into the market.

a junior developer with some social skills can easily outperform a senior engineer

given the fact that the code generation part is mostly solved, the bottleneck is getting to know people with businesses that need software.

selling software is a different skill than making it

AI is not gonna sell software for you, at least yet, but it can now be your 20/month employee that writes all of the code

[OBEler]

10 minutes ago, PolyPeter said:

guys!!!! for fucks sake!!! use your imagination, this is just another technological revolution, new jobs will be created

Nope. Not this time. Maybe as an AI trainer. But this is just a short phase to make Ai even more powerful. 

 

Edited Friday at 06:41 PM by OBEler

[bazera]

18 minutes ago, OBEler said:

No software engineer then necessary. 

What's your plan? You're gonna transition into a different industry, or?

[OBEler]

@bazera no. I retire

[pablo_aka_god]

I would say do it if it's your purpose in life and you truly love it!

If you are in for the money don't do it, that ship has already sailed and it's much harder today to make money in IT.

Possible but as hard as making money in any other profession, only the top 5% make it.

[Thought Art]

AI is EXTREMELY energy intensive which I suspect is part of the drive for building them. Nuclear and Oil companies are salivating, This also means it’s likely going to be really expensive and I doubt it will become as good as humans. So, I think people are safe.

[something_else]

20 hours ago, OBEler said:

@something_else

"it can do the work of two or three". 

Right now. Wait two years and it can do work if a whole team. 

No software engineer then necessary. 

 

 

 

 

Dude right now it can't even tell me whether we should use a library or build an implementation from scratch for a particular project at work. It has access to our entire codebase and knowledge base, our coding standards, and our project requirements. All the context it could reasonably need, and it still flipfops back and forth between recommending both of them after every prompt. The slightest bit of pushback and it does a 180. Slightest bit of pushback again and it does another 180. Point being it is incapable of making decisive architectural decisions like this reliably.

For the same project I vibe coded a POC in like 30 minutes for a project that is going to take us around 3 months to build properly. Amazing? Yes. Except we gave the POC to our designer so she could vibe code some other features into it for experimentation and it just fell apart the more changes she made. It was a house of cards. Pure vibe coding is very unsustainable because the poor architectural choices of the AI compound exponentially over time.

Imagine somebody non-technical came to you tomorrow and asked you to work on their web app that they'd built entirely with Claude vibe-coding over the past year. Do you seriously think that has any chance in hell of being remotely maintainable?

The idea that we would let LLM agents operate autonomously in place of an entire dev team is kind of a joke, and I don't see that changing in the next few years. Maybe in 5-10.

 

Edited yesterday at 03:15 PM by something_else

[Valach]

Senior Software engineer here, It is gonna make it harder to enter the field or start your own business (unless your idea is super unique), but people with experience are not getting replaced soon.

Edited yesterday at 03:45 PM by Valach

[bazera]

 

[Joshe]

1 hour ago, bazera said:

 

Jesus. I didn't know Opus 5.5 was such a big leap. This levels the fuck out of the playing field, which kind of sucks, lol. 

[pablo_aka_god]

When will this tech reach a ceiling?

if the current improvement trend continues for 5 more years 70% of all jobs are automatable.

If it plateaus then jobs are safe till next leap in AI.

I think most jobs we do today are automatable. 

We need to create a new marketplace for new types of jobs like research oriented, creative, etc.

[LastThursday]

Capitalism will take the cheapest path. AI is not cheap to run and maintain and those costs will eventually trickle down to the users of AI, at the moment it is probably underpriced to get people hooked on it. If AI ends up costing more than a human to employ for the same level of productivity then humans will win over AI. Will the AI cost problem be solved? Only by ditching LLMs, which are the core of coding AIs at the moment. But I don't think anyone really has new ideas about what would be cheaper and as powerful as LLMs to run.

The other side of the coin is that if you're going to rely on AI for all commercial coding practices, then it really has to be rock solid especially when it comes to the security side of things. I reckon you'll always need human experts in the loop to ensure you're getting good software out - those jobs won't disappear.

The most likely scenario is that there will still be coders, but they will be more like "prompt engineers" where the AI does all the grunt work, and the coders have all the ideas. This is neither good nor bad, just different. But, there will still be jobs.

[bazera]

4 minutes ago, LastThursday said:

The most likely scenario is that there will still be coders, but they will be more like "prompt engineers" where the AI does all the grunt work, and the coders have all the ideas. This is neither good nor bad, just different. But, there will still be jobs.

That is already the case.

[LastThursday]

32 minutes ago, bazera said:

That is already the case.

For sure, but it's not universal yet.

[Heaven]

I believe anything to do with coding is most likely be gone in the next 5 years but of course new job opportunities will rise.

@LastThursday Employees are 100x more expensive than ai/robots in the West.

[Leo Gura]

https://www.instagram.com/reel/DcL_vVMCC2u/