Have you lost your job to AI?

Have you lost your job to AI?

  • Yes, and I've been struggling to find work since

    Votes: 5 7.5%

  • Yes, and I've landed a new job already

    Votes: 0 0.0%

  • No, but I'm expecting it

    Votes: 19 28.4%

  • No, and I'm pretty confident my job is safe

    Votes: 43 64.2%
  • Total voters
    67
Pegging said:
I have seen similar.

In particular, people in BI or reporting are being replaced by AI.
I still think you need someone who understands data though, because AI is not always clever at recognising crap.

People doing data recons can also have a lot of their work automated.

Just in those two spaces we can get work that took 500 hours a month before down to an hour.
Click to expand...
I work in this space and manage a small team. We don’t “employ” agents but use AI to fast-track projects.
All it has done is made us very efficient. It also means we are taking on more and more work from poorer performing regions without our department collapsing under the pressure. Some people may eventually lose their jobs but it won’t be us.

Agree on the human aspect. AI produces some really bad insights in the wrong hands. On top of that, when these smart arses produce these numbers(outside of our department) they don’t actually implement any repeatable methods. All they are doing is mashing together numbers, which are quite often wrong.

Agents are also known to lie. And even when you add a spy, they’ll band together and find a way to continue the lie. Quite human in that respect.
 
Sinbad said:
It's only cheaper because it's completely unsustainable.

Anthropic just posted a half year loss of $80bn.

They will want to start recouping these costs soon. Claude price ++++++++
Click to expand...
It is going to run aground for the claude/codex subscriptions at some point because the subscriptions are heavily subsidised by VC and other "AI ecosystem incest deals".

At some point AI as a subscription is going to either become very slow/crappy or super expensive. It basically cannot continue like it is indefinitely.

That being said, some big places are buying their own multi million rand DGX clusters and setting up local, on premises (on in DC) LLMs. So they would be able to continue, but hopefully they would rather augment their staff than replace them. We shouldn't have non-technical CEOs out there going "claude build me this epic tool and place it on the web. Oh ja... make no mistakes" (I know people who verbatim quote that last part as if it's a magic enchantment. lol)
 
When i started my business some guys approached me to make an app for my systems, got it done and they didnt charge me as it was a development they wanted to sell to other companies in the same space.
They then wanted to charge me quite a packet once it was ready for the market.
I declined.
I went onto Claude and designed the same, no wait, better app for free, in 20 minutes.
All these jobs are going to be redundant soon, Ai can do it cheaper, faster, better and it just doesnt make sense to hire people who in today's times are more problems than they are worth. I use ChatGPT daily for stuff, paid subscription, same with Claude now, the costs I pay per month are worth it for me. For others it may not be, same as me having a 500mb line at home, sure most people dont need it, neither do I probably but hey I want it.
 
ToxicBunny said:
but relying on it to do the same thing 100000 million times without fail and getting it right every time is not something I think its good at.
Click to expand...
You've touched on a point many people using (and talking about) "AI" don't understand.

AI is "non deterministic" which means it is likely to do those 100000 million iterations differently.

If you ask an LLM "what is one plus plus one" it COULD technically answer with
"two"
"oneone"
"eleven"
"bicycle"

Stupid example I know, but it's for illustrative purposes. It also doesn't "understand" or "think" (even though that baas ted will happily declare "thinking" while it's processing).

I'm not a fan of "agentic everything" and some people want to do away with APIs and just have an "Agentic access point" where people can send instructions to an agent...
 
PaulKempTheSecond said:
I work in this space and manage a small team. We don’t “employ” agents but use AI to fast-track projects.
All it has done is made us very efficient. It also means we are taking on more and more work from poorer performing regions without our department collapsing under the pressure. Some people may eventually lose their jobs but it won’t be us.

Agree on the human aspect. AI produces some really bad insights in the wrong hands. On top of that, when these smart arses produce these numbers(outside of our department) they don’t actually implement any repeatable methods. All they are doing is mashing together numbers, which are quite often wrong.

Agents are also known to lie. And even when you add a spy, they’ll band together and find a way to continue the lie. Quite human in that respect.
Click to expand...
If you can get AI to help you do more and pay for itself that is great. If you need to let people go in order to pay for AI then you’re probably better off without it.
 
warrenpridgeon said:
You've touched on a point many people using (and talking about) "AI" don't understand.

AI is "non deterministic" which means it is likely to do those 100000 million iterations differently.

If you ask an LLM "what is one plus plus one" it COULD technically answer with
"two"
"oneone"
"eleven"
"bicycle"

Stupid example I know, but it's for illustrative purposes. It also doesn't "understand" or "think" (even though that baas ted will happily declare "thinking" while it's processing).

I'm not a fan of "agentic everything" and some people want to do away with APIs and just have an "Agentic access point" where people can send instructions to an agent...
Click to expand...
Yes it is far from deterministic.
One prompt... the same prompt, has given me 10 different variations of code, many of which were flawed.
Yet, my company's bosses think this is quite OK, especially when they're handing the reins of critical code that should not be messed with, to AI. But it's not AI's fault you see, no sir, it's these "overpaid, lazy developers". That's the kind of toxic culture I work in.

And I even see it in the artworks I sometimes generate to troll Mr. B with. Most of the time, to get it to be deterministic I have to input something, I am finding that all that works that approaches deterministic output is to feed the AI an already fleshed out partial sketch and then I have to go to pains to get it to recognize it MAY NOT change certain things.

It cuts down the time, but does NOT an CANNOT replace my creativity or ingenuity.
The modern world prizes quantity over quality and counting coins... there's no place for creativity or ingenuity anymore.

As mentioned, graphic designers are on the chopping block. I am helping someone who got laid off from those delightful people known as CLICKS, their head DTP/Graphic Designer, who now has to sleep on people's couches because they've lost EVERYTHING. CLICKS reckons AI can do it all for them. So now I make a point of looking for mistakes in their AI slop pamphlets... so when I see a mistake I will grind them on each point, as is my right under the CPA.
 
EmmanuelGoldstein said:
Yes it is far from deterministic.
One prompt... the same prompt, has given me 10 different variations of code, many of which were flawed.
Yet, my company's bosses think this is quite OK, especially when they're handing the reins of critical code that should not be messed with, to AI. But it's not AI's fault you see, no sir, it's these "overpaid, lazy developers". That's the kind of toxic culture I work in.

And I even see it in the artworks I sometimes generate to troll Mr. B with. Most of the time, to get it to be deterministic I have to input something, I am finding that all that works that approaches deterministic output is to feed the AI an already fleshed out partial sketch and then I have to go to pains to get it to recognize it MAY NOT change certain things.

It cuts down the time, but does NOT an CANNOT replace my creativity or ingenuity.
The modern world prizes quantity over quality and counting coins... there's no place for creativity or ingenuity anymore.

As mentioned, graphic designers are on the chopping block. I am helping someone who got laid off from those delightful people known as CLICKS, their head DTP/Graphic Designer, who now has to sleep on people's couches because they've lost EVERYTHING. CLICKS reckons AI can do it all for them. So now I make a point of looking for mistakes in their AI slop pamphlets... so when I see a mistake I will grind them on each point, as is my right under the CPA.
Click to expand...
Some people take entirely the wrong approach (many argue that there is no "right" way to use AI).

I like it for optimising mundane stuff... I built a template for social media posts. I write all the content and then I say "bot go take quotes from my ramblings and put them on cards" to post on social media.

For doing CRUD stuff it's pretty ok if you've got an existing structured system (and it actually listens)... if you aren't paying attention you will end up with inline SQL inside controller methods - because whatever stackoverflow post or public github "how to get started with C#" demonstrated that for brevity... lol.

It will HAPPILY run off and spend all the magic tokens on building ENTIRELY the wrong thing. You have to beat it with a stick sometimes (not too much or they block your claude account now... lol clanker lives matter)
 
EmmanuelGoldstein said:
Yes it is far from deterministic.
One prompt... the same prompt, has given me 10 different variations of code, many of which were flawed.
Yet, my company's bosses think this is quite OK, especially when they're handing the reins of critical code that should not be messed with, to AI. But it's not AI's fault you see, no sir, it's these "overpaid, lazy developers". That's the kind of toxic culture I work in.

And I even see it in the artworks I sometimes generate to troll Mr. B with. Most of the time, to get it to be deterministic I have to input something, I am finding that all that works that approaches deterministic output is to feed the AI an already fleshed out partial sketch and then I have to go to pains to get it to recognize it MAY NOT change certain things.

It cuts down the time, but does NOT an CANNOT replace my creativity or ingenuity.
The modern world prizes quantity over quality and counting coins... there's no place for creativity or ingenuity anymore.

As mentioned, graphic designers are on the chopping block. I am helping someone who got laid off from those delightful people known as CLICKS, their head DTP/Graphic Designer, who now has to sleep on people's couches because they've lost EVERYTHING. CLICKS reckons AI can do it all for them. So now I make a point of looking for mistakes in their AI slop pamphlets... so when I see a mistake I will grind them on each point, as is my right under the CPA.
Click to expand...
That's tough man,

Just my own opinion here, I would be a bit more hesitant to say that AI cannot and will not be able to do this or that. That's like saying that a 4. year old kid cannot and will not be able to solve complex mathematical problems.....there's a distinct possibilty that it's only a matter of time.
 
Alexander the Straight said:
That's tough man,

Just my own opinion here, I would be a bit more hesitant to say that AI cannot and will not be able to do this or that. That's like saying that a 4. year old kid cannot and will not be able to solve complex mathematical problems.....there's a distinct possibilty that it's only a matter of time.
Click to expand...
And a significant change in the underlying tech. We are still some distance from AGI.
 
Alexander the Straight said:
That's tough man,

Just my own opinion here, I would be a bit more hesitant to say that AI cannot and will not be able to do this or that. That's like saying that a 4. year old kid cannot and will not be able to solve complex mathematical problems.....there's a distinct possibilty that it's only a matter of time.
Click to expand...
The psychiatrist tells me that human cognitive processes, and sentience would not in the short and medium term be artificially generated.
 
Development is a very slippery slope. Manually typing code , scaffolding and a lot of what I call grunt work is gone, Agents handle all that easily. Overall architecture, direction and validation of logic, those are still massive holes.

I can comfortably run multiple agents working on whatever and then review and work with them from a logic perspective and keep an eye on conformance etc but I also have folks who are trying to have AI do that as well as it still a serious problem. It goes down tangents, over complexifies, ignores already walked and prepared structures to implement something new and often just misses the point. I can't say its just the models simply because I use the same. The difference is people leaning on years of experience in validating or vetting it and often their frustration in "why doesn't it just work or do that", which is a very common refrain from most devs. Development, is a hugely frustrating job and a lot of trial and error as well as moving goal posts. A laymen expecting it to work perfectly is deluding themselves, unknowingly because they are also ego pumped by the models.

Any way, from a development point of view, AI should make you extremely more efficient so yes, I would expect companies with large cohorts of developers to start downsizing. My concern for the industry, is that we are not slowing down but CUTTING off junior ingestion and mentoring. Its effective now because many have come up and had to train a critical thinking muscle that I say is lacking in the younger generations and with AI is being reduced even further. THAT is a big worry of mine, as a person in the industry and parent.
 
Archvile said:
The company where I worked, in the fintech space, as a senior full-stack software engineer, took an aggressive AI adoption stance as a means to keep up with the trends. Their multi-pronged strategy involved both using AI in our daily work and integrating it into our products.

After a few years everyone started submitting pull requests for code changes - our business analysts and even the CEO got in on the action.

18 months later I received a letter of termination saying my role is redundant and they can't afford me any longer. 11 years at the company as employee #1, and I got replaced by an AI because it was cheaper than my salary.

Do you have a similar experience, or fear your job is at risk? Share your stories, I'd love to hear them.
Click to expand...
Sorry to hear man. Its an incredibly tough spot to be.

In your opinion, do they have both the staff and skills to replace you post retrenchment?
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X