The 2025 Bum Thread

[D.P.]

It's always strange to here people say this to me, because I see the complete opposite, but I don't know what context most people use it. But people obviously have success, so maybe I need to temper my expectations with what I use it for, which isn't the super popular stuff like Javascript/Python/Typscript/Node etc.

I'm primarily in automation, Ansible, Terraform, Bash and at home for home-projects sh and C. Terraform isn't too bad in LLMs but the rest is absolute gash. I find they give badly structured, out of date code running with parameters which don't exist, I damn near got into an argument with one as it started making up parameters even after I spelled out which were available. I'd dread to think of the Javascript/Pyhton code they provide considering how much bad code for those is on the web to be scraped.

I did end up asking Claude where the hell it was getting it's info, because it didn't have a clue

The more I try to use them, and see other people try to use them the more suspect I get of the claims of big companies pushing their agenda. I'll be kept in a job fixing AI slop if nothing else.

I was at an applied ML conference and one of the sessions was dedicated to AI in software development. Pretty much the consensus from every expert was that it will provide small efficiency gains just like every other avancement before that and while that might allow minor staff reductions in some companies , the efficiency gains will be offset by developing either more projects or more complex projects. If all your competitors gain efficiency from AI, you cannot leverage AI you get an advantage, merely keep up. As such AI might increase demand for software developers.


There were also industry leaders saying they stopped Copilot licences because the ROI was not there

 

[CREATIVE!11]

I don't understand your opinion on trades and physical jobs, AI can't do any of them, and there is always great demand. Even when (for example) the building industry crashes, tanks home maintenance, is always ongoing.
Art and creative industries are already swamped with AI. Most of it terrible. But then most creative industries are utterly undervalued and its impossible to make a living in the majority of them.
Seeing a lot of AI in software engineering. Its all around the fringes, no code solutions, nibbling all around the edges of it, and people using it to do donkey code.
Won't replace the high end, but you can't get to the high end career wise if theres no low end as a stepping stone. That has long been an issue in IT and its going to get worse.

This is true, as a junior still just under 2yoe, You need constant long term exposure to grow, if you don't get it, you can't get to high end.

 

[OspreyO]

There is a high level of demand compared to supply today, but supply is easily increased and demand wont increase at the same rate. In fact some would be highly dependent on the economy. Maybe people will eat out less, and not pay for gardeners or house upgrades.



You answered yourself here. AI art is terrible, and people will always value human art over AI just like they value real diamonds over manmade even when the latter are superior.


I think art valuation is always highly variable. Of course, impossible gor most to make a living as you say.



As i said, AI in software will only be another efficiency improvement, a tool that helps developers create more complex software.

I think you're mistaken. Trades take a long time to acquire as long if not longer than degrees. Never mind retaining good people. The industry has been struggling with a massive skills shortage in trades and manual labour jobs for years.

I think theres a confirmation bias here assuming only only software will remain unaffected. When everything points to that being one of the first affected. IT people even on these forums describing layoffs and slacking of demand for all except the top people. That's before AI has really got going.

 

[OspreyO]

I was at an applied ML conference and one of the sessions was dedicated to AI in software development. Pretty much the consensus from every expert was that it will provide small efficiency gains just like every other avancement before that and while that might allow minor staff reductions in some companies , the efficiency gains will be offset by developing either more projects or more complex projects. If all your competitors gain efficiency from AI, you cannot leverage AI you get an advantage, merely keep up. As such AI might increase demand for software developers.


There were also industry leaders saying they stopped Copilot licences because the ROI was not there

I think you have to have a business dataset that can leverage AI. You also have to have the technical ability to leverage it. You've got to have the business desire to invest the resources into it. Often people are cheaper. I've met the same resistance to automation.

Of course some places will embrace it.

 

[Ocelot]

Despite all the gloom and doom about "AI is going to take our jobs" when AI is still in its infancy and we are seeing the cracks already. Software Engineers are not loosing their jobs to AI, they are loosing their jobs to cheaper countries where they are being outsourced.

That's your biggest threat and current problem.

My last role we had about 10 software devs in our team. The 8 of them were in India and Poland. There are looking for more devs in those countries but not locally.

In the late 80s/early 90s CASE was going to replace programmers. That never happened.

 

[malachi]

In the late 80s/early 90s CASE was going to replace programmers. That never happened.

Yep, here we are STILL waiting..............

 

[NickK]

Yep, here we are STILL waiting..............

Scoobydoo ending - unmasks AI to reveal Excel.

 

[Demon]

In the late 80s/early 90s CASE was going to replace programmers. That never happened.

I started my Career when CASE was a thing and saw some good implementations, but it died off because 99% of all software engineers don't have the capacity to think up front and must hack around doing it until it roughly works, then throw it over a fence to a test team to spend weeks finding all the problems and leaving many timebombs just sitting around..

I don't advocate for CASE tools, I am simply saying they are just incompatible with the reality of the workforce it was being applied to.. the downside to trying to be that structured is it kills all creativity which just turns your brain off and you go and seek something more interesting..

I'm naturally driven to create highly abstract designs to ensure I only do the boring bits once, and then spend the rest of the time focussed on the more creative/unique stuff, so there are some lessons from CASE I've carried forward..

 

[NickK]

Well I’ve entered myself into an AI hackathon in April. Nothing like pushing yourself.

 

[malachi]

Scoobydoo ending - unmasks AI to reveal Excel.

Second episode - unmasks Microsoft Copilot to be Clippy!

 

[malachi]

Well I’ve entered myself into an AI hackathon in April. Nothing like pushing yourself.

Nice!

 

[D.P.]

I think you're mistaken. Trades take a long time to acquire as long if not longer than degrees. Never mind retaining good people. The industry has been struggling with a massive skills shortage in trades and manual labour jobs for years.

Some trades, but you you can start as an apprentice right away. And No one really cares about a fresh Comp Sci graduate, the value is in people who many years experience.
Plus mid_gen mentioned pother jobs like Gardener and hotelier industry. Nothing stops you starting your own gardening busy for example, of becoming a waiter.

Some trades have a reduced supply of tradesmen, but there is not a big barrier to entry. It is merely that people don't want to be a leccy etc., and things like Brexit has driven out some tradesmen. Software engineering requires a certain level of intelligence, problem solving ability and skill. which reduced the total pool of applicants. Not saying some trades are not skilled, but the overall skill level and abilities are smaller as the set of problems is quite restricted so you can quickly perfect the knowledge aspects.

if suddenly a lot of jobs disappeared to AI like software engineers, nothing is really stopping these people picking up a trades job. ironically, my Dad stopped being a Lecturer and became a landscape gardener!

I think theres a confirmation bias here assuming only only software will remain unaffected. When everything points to that being one of the first affected. IT people even on these forums describing layoffs and slacking of demand for all except the top people. That's before AI has really got going.

I see precisely zero signs of software engineering being affected directly by AI taking over roles. As others have dais, the main drive to layoffs is outsourcing. Also, although many big tech companies are laying off developers, this is related to over hiring in covid and a weird land-grab over developers hiring anyone with minimal experiencing and driving up ridiculous salaries. Demand for software engineers is actually quite high but many sectors are also controlling budgets. Indirectly some layoffs have happened due to AI, but it isn't AI replacing human coders, merely companies wanting to invest in AI so laying off other staff so they can hire AI/ML engineers etc.

 

[mid_gen]

It's always strange to here people say this to me, because I see the complete opposite, but I don't know what context most people use it. But people obviously have success, so maybe I need to temper my expectations with what I use it for, which isn't the super popular stuff like Javascript/Python/Typscript/Node etc.

I'm primarily in automation, Ansible, Terraform, Bash and at home for home-projects sh and C. Terraform isn't too bad in LLMs but the rest is absolute gash. I find they give badly structured, out of date code running with parameters which don't exist, I damn near got into an argument with one as it started making up parameters even after I spelled out which were available. I'd dread to think of the Javascript/Pyhton code they provide considering how much bad code for those is on the web to be scraped.

I did end up asking Claude where the hell it was getting it's info, because it didn't have a clue

The more I try to use them, and see other people try to use them the more suspect I get of the claims of big companies pushing their agenda. I'll be kept in a job fixing AI slop if nothing else.

I run LLMs locally, so no cost to me (other than a little electric i guess). Even 1.5billion parameter coding models provide plenty of useful autocomplete, I'm working in C++ and the main models all know Unreal.

There's quite a bit of boilerplate generally in gameplay code, big timesaver.

The value AI provides doesn't come from 'hey AI make a game for me', it comes from 'you just typed the function named algorithmXYZ, is this the implementation you were about to write? Yep, done.

Not replacing the senior, experienced programmers, but streamlining our workflow. Lots of value there.

 

[OspreyO]

Some trades, but you you can start as an apprentice right away. And No one really cares about a fresh Comp Sci graduate, the value is in people who many years experience.
Plus mid_gen mentioned pother jobs like Gardener and hotelier industry. Nothing stops you starting your own gardening busy for example, of becoming a waiter.

Some trades have a reduced supply of tradesmen, but there is not a big barrier to entry. It is merely that people don't want to be a leccy etc., and things like Brexit has driven out some tradesmen. Software engineering requires a certain level of intelligence, problem solving ability and skill. which reduced the total pool of applicants. Not saying some trades are not skilled, but the overall skill level and abilities are smaller as the set of problems is quite restricted so you can quickly perfect the knowledge aspects.

if suddenly a lot of jobs disappeared to AI like software engineers, nothing is really stopping these people picking up a trades job. ironically, my Dad stopped being a Lecturer and became a landscape gardener!



I see precisely zero signs of software engineering being affected directly by AI taking over roles. As others have dais, the main drive to layoffs is outsourcing. Also, although many big tech companies are laying off developers, this is related to over hiring in covid and a weird land-grab over developers hiring anyone with minimal experiencing and driving up ridiculous salaries. Demand for software engineers is actually quite high but many sectors are also controlling budgets. Indirectly some layoffs have happened due to AI, but it isn't AI replacing human coders, merely companies wanting to invest in AI so laying off other staff so they can hire AI/ML engineers etc.

A just started apprentice is not tradesman. That's like saying a freshers week comp sci undergraduate is a developer.

Reminds of this....

 

[OspreyO]

That said I think the impact of AI is way over stated.

 

[malachi]

That said I think the impact of AI is way over stated.

It is, too many people get excited over something which has been around for 5 minutes.

We all have to get up and go to work for the foreseeable future. That won't change.

 

[NickK]

I run LLMs locally, so no cost to me (other than a little electric i guess). Even 1.5billion parameter coding models provide plenty of useful autocomplete, I'm working in C++ and the main models all know Unreal.

There's quite a bit of boilerplate generally in gameplay code, big timesaver.

The value AI provides doesn't come from 'hey AI make a game for me', it comes from 'you just typed the function named algorithmXYZ, is this the implementation you were about to write? Yep, done.

Not replacing the senior, experienced programmers, but streamlining our workflow. Lots of value there.

The issue for many applications is the lack of security in the resulting code.

Amazon Q Developer uses the comment description of the function to generate the code which is quite nice as it forces a little more scoping.

Now that simply means more reliance on security inspection tooling in the pipeline etc, code reviews etc as normal but the pressure to remove experienced engineers means those tools and pipelines end up being either unused or neglected at the wider cost.

Anyways - two job applications and I'm on to thinking about the AI hackathon ideas (it's open format which is good for rusty coders like me).

 

[kaiowas]

We all have to get up and go to work for the foreseeable future. That won't change.

Excuse me, but this is the “bums” thread. “Getting up and going to work” is down the hall.

Been a bum since mid January now (Took VR when it was offered, had been there for many years so got enough of a payout to keep me off the streets for a while). Was working in IT infrastructure but not convinced I want to stay in that line of work or do something different. I’m also starting to wonder how I ever found the time for a job, been plenty busy around the house/garage.

 

[malachi]

Excuse me, but this is the “bums” thread. “Getting up and going to work” is down the hall.

Even Bums have to get up and work at some point lol

Well, my last weekend of being a Bum. Will officially be back in employment starting 1st April.

Since I walked out of my job on 7th February, its been a good run

Gran Canaria
Tenerife
Mexico
France
UK (friends and family)
Egypt

Are all places I have visited during my Bum era. Ready for work next week......I hope!

 

[mid_gen]

I'm doing two days a week at a uni, which will be running until early May.

The other three days I'm alternating between going out filming climbing, or working on my Unreal projects to spruce up the portfolio, weather dependent.

My Youtube channel got monetised yesterday, so that's been going well at least!

More chats coming soon about future work, still likely to just go freelance and start working day rates.