After 18yrs & multiple bullets dodged..

[-]

Wonder_Weenis@reddit

I have yet to see this actually work.

What effin model you got over there, replaced an entire team of people?

[-]

Calleb_III@reddit

It won’t work. It will have the appearance of working for a while, as AI is very good and appearances. Execs will get their fat bonuses While under the hood everything will crumble to dust.

In 6-12 months it will become apparent and Manglement will hire some “AI shepherds” in a low cost centre and give it another 6-12 months. Just enough for execs to collect their bonuses.

After that it will be such a mess that the execs will just outsource it (and collect their bonuses)

[-]

ZippySLC@reddit

Given the constant bad level of decision making I've seen from execs over my nearly 30 year career in tech I think replacing them with an AI would likely be a better move

[-]

Sh1rvallah@reddit

Need to replace the share holders first. All this shit comes down to short term investing ruining everything.

[-]

Vietnamst2@reddit

Shareholders want their money. If AI provided sure income for them, they would gladly accept it as new board member and fire everyone else.

[-]

lerun@reddit

I see there is a lot of collecting bonuses in there

[-]

The people responsible for causing this train wreck always get rewarded for fixing their own cocks ups because it looks like cost savings and often that's all that matters. Even if, in reality, this actually costed the business productivity and profits but they've cleaned up the numbers so impact isn't as obvious as it should be.

[-]

Hebrewhammer8d8@reddit

Money manipulation at the finest?

[-]

lerun@reddit

Strange how this works when you are the one controlling the information flow

[-]

ScreamOfVengeance@reddit

Bonuses are the main objective at that level

[-]

plumbumplumbumbum@reddit

What else are they going to do with al the money the saved on the eliminated teams payroll?

[-]

dustojnikhummer@reddit

Parachutes and bankrupcy

[-]

Cormacolinde@reddit

Heck, the AI will LIE and GASLIGHT people about its shit working. It happens all the time.

The website isn’t working

AI: The website is working perfectly fine

I can’t get to it!

AI: It must be your fault, or your phone is broken

And meanwhile, the AI is deleting the prod environment to cover its tracks.

[-]

Mindestiny@reddit

I mean... That dialogue sounds like your typical "the internet is broken" user ticket lol.

No, Sandy, the internet is not broken. Google is not down. You just didn't connect your laptop to the Wi-Fi network.

[-]

Cormacolinde@reddit

Exactly. Because the AI learned from those jokes and funny videos and bash.org snippets. And it doesn’t understand the difference between jokes and professional behavior, it has no sense of ethics or conscience.

[-]

Mindestiny@reddit

That's not gaslighting and telling lies though, and certainly not "deleting prod to cover its tracks."

A response not being in a professional tone is vastly different than assuming intentional malice where there is none. I'm not gonna sit here and argue with another person who's clearly feverishly anti-AI, I was merely pointing out that you used a pretty terrible example if you were going to claim it's maliciously telling lies when that response is the correct response to 99% of those reports.

[-]

Cormacolinde@reddit

There’s no intention. An LLM simply imitates, reproduces likely scenarios and text it has been trained on. But its lack of understanding or intent is the problem.

[-]

Mindestiny@reddit

Except that's not what you claimed.

You claimed it's intentionally telling lies and gaslighting people, then covering its tracks.

That cannot be true without intent and understanding. So you're just tossing hot air.

[-]

Cormacolinde@reddit

If we toss dice and get a 1 five times in a row, we blame the dice. There is no intention on the part of the dice, yet the 1s were still tossed.

Likewise, we can say an LLM lies, without any intention being part of it. An LLM can produce output that reads “cats have five legs”. The intent is not relevant, it’s still false.

An LLM is nothing more than a fancy dice thrower, it just sounds more coherent. The correct terms would probably be something like “produces statements with negative truth-value” but such verbiage makes communication heavier without really helping understanding.

[-]

Mindestiny@reddit

Likewise, we can say an LLM lies,

No, we literally cannot. If you're going to sit here and make a silly semantic argument, you might want to have the definition of the word correct.

noun

noun: lie; plural noun: lies

an intentionally false statement.

As you said, there's no intent, so it cannot lie. It can accidentally give you incorrect information, but it 1000% cannot "lie." It cannot gaslight someone either, which also explicitly requires intent.

An LLM is nothing more than a fancy dice thrower, it just sounds more coherent.

You're really not understanding what an LLM is and how the brokerage applications that are built on top of them function if that's your bottom line. It's not a "fancy dice thrower," it's not randomly giving output. Like on a fundamental level that statement is just... so, so completely incorrect.

The correct terms would probably be something like “produces statements with negative truth-value” but such verbiage makes communication heavier without really helping understanding.

The correct term is "it gave incorrect information." or in more laymen's terminology it "hallucinated." Whatever you just spit out to try to move the goalposts is total nonsense. And that's where I'm leaving this.

[-]

robotbeatrally@reddit

So GPT kept responding to me in Arabic yesterday (I have never spoken or used Arabic in any of my history with it) and I asked it why are you responding in arabic? It said oh sorry its an artifact yadda yadda. here's a better answer (only to respond in arabic like 4x) Eventually I got it to stop responding in Arabic. It was the weirdest glitch I've experienced with AI using a normal ass pretty clear and specific simple prompt so far, and I wanted to share with someone but have no friends. So that's my story.

[-]

Funlovinghater@reddit

Wtf are bonuses and why am I not getting them?!

[-]

davy_crockett_slayer@reddit

Harness https://www.harness.io/

[-]

bv728@reddit

Pretty much nobody is actually doing layoffs because of AI at this point. We're in the middle of a massive recession that nobody wants to admit is a recession and they're all reaching for layoffs as a result.
It also bolsters the case for more AI spending, keeping the bubble going that helps cover up the recession, and execs are either genuinely convinced that AGI is six months away (like it has been the whole time), or trying to keep their portfolio growing for another year.

[-]

cplusequals@reddit

Unemployment is relatively low and has been dropping this year. AI might be a bubble for the S&P in terms of valuation, but not for GDP which actually does rely on tangible production. The only major economic stressor right now is the price of oil, but $100 a barrel isn't the red warning light it used to be after the dollar devalued 30% post COVID. Just to put it into perspective, gas prices right now are nationally hovering around $4. Using 2026 dollars, we were at ~$6.50 for a few months back in 2022.

We were more likely to enter into a recession at this point last year than we are now.

[-]

kariam_24@reddit

GDP by what? By planning data center and inflating prices of ram, cpu, gpus while money circulate between Nvidia, Amd, Intel and cloud, AI companies? While cutting headcount is done to fund more of AI in hopes that it will be profitable instead it allowing to be more productive?

[-]

cplusequals@reddit

Data centers are not even a minor driver of GDP and the vast majority of processing power is overwhelmingly productive and not trash AI. Inflated prices due to high demand for electronics that are imported actually count against GDP. Remember -- the D stands for domestic. Pretty much all of the domestic growth (construction, domestic infra) is going to be offset entirely if not completely eclipsed by the tech imports that go into it.

You can't really talk about "headcutting" while unemployment is in a really good spot right now. I'm pretty sure demand for IT jobs has increased alongside excitement from AI. Layoffs are up but so are new hires. There's a lot of churn but not widespread downsizing (even if the headlines always print when a major company does it).

I'm personally convinced we're going to have a larger IT sector in the next decade not a smaller one.

[-]

kariam_24@reddit

Fake GDP numbers of orders going around same companies like i mentioned. Russia also have GDP growth in theory.

How unemployment is in good spot when there are layoffs of thousands employes like recent 30k at Oracle or earlier Meta and other companies?

Sure there is more hiring in India.

Not sure why are you contradicting yourself, praising AI bubble.

[-]

cplusequals@reddit

OK, so magic stats aren't real. That's... well, not much point in talking.

You're fairly stupid if you think me pointing out AI data centers are not driving GDP growth is a contradiction.

And the numbers are what the numbers are. 30k people laid off at once? Seasonally adjusted employment dropped net 330k going into April, so maybe you can use that as a reference for why Oracle alone isn't driving the unemployment rate up.

I think you're done here. Toodles.

[-]

Library_IT_guy@reddit

CEO fires the people that make the company run and replace it with AI because they drank the kool-aid.

CEO gets big bonus

Company starts to die over time due to inferior product/support, but makes massive short term profits.

CEO moves to new company.

Company dies or is a shadow of itself due to losing quality after the process, but by now the CEO is already in a new position doing the same shit.

A tale as old as time, just replace AI with outsourcing to 3rd world countries.

[-]

bruhgubgub@reddit

When AI goes from "artificial intelligence" to "actually Indian"

[-]

TerrorBite@reddit

Wait until you find out about https://indiaai.gov.in/

[-]

discosoc@reddit

Was en entire team actually replaced? AI helps downsize, often drastically, but the OP doesn’t mention the entire team being replaced.

[-]

kariam_24@reddit

Downsizing is pretext to transfer more funds into AI hoping it will be profitable, it is same like mass outsourcing, cutting cost while lowering support or development quality, for sake of shareholders or executives bonuses during that quarter.

[-]

synthetix808@reddit (OP)

Due to the way large corps do it, i have no idea if other members of the same team were effected. However, I do know it was roughly 5% of the several thousand employees. So, possible??. The company had pushed AI usage/dev to the point were team members were writing code to affect the other team members roles and duties. So they are/were definitely betting the house on it.

[-]

DrockByte@reddit

It's the same mindset as, "our mechanics were losing us money, so we fired the whole department and have put a stop to all vehicle maintenance for the entire fleet. Quarterly profits are through the roof!"

[-]

synthetix808@reddit (OP)

yep. such a shortsighted mindset. but here we are

[-]

Ok_Tap7102@reddit

Opus 4.6, the fact people don't know this is why they're disappearing

[-]

Khue@reddit

It won't. It's a gamba to promote short term profits over long term stability of an organization. Eliminating positions instantly bumps profit margins.

[-]

randomlyme@reddit

I’m a low level exec running AI enablement, leading a multi national swat engineering team. I was just told my position was eliminated. I at least have a couple of months notice. It’ll be interesting to me to see if they can even tell about the value I add or will it be missed at all. I’ve got enough contacts to hear how Thu gs are going after I leave but it’s more morbid curiosity.

Regardless, I’m updating linked in and activating my network and figuring I’m gone (hopefully) after the bonus lands. I’m grateful for the long run but these are precarious times for all! Best of luck brother!

[-]

flurfdooker@reddit

I think the only thing saving my job at this point is that I'm one of three people with physical access to our AI cluster. It requires clearance to access, and no one else lives in the area. Also, one of our "engineers" yesterday tried to image a DGX workstation by copying an ISO file to an NTSF formatted USB key and expected it to work. I asked them to show it to me and they plugged it into their laptop and said: "It's right there. DGXOS-XXX-XXX.iso"

At some point this will become a million dollar talent.

[-]

rubmahbelly@reddit

Wow.

[-]

lectos1977@reddit

Yeah, I am being pushed to do AI for everything. I asked how AI coils would run network cables and be adaptable to anything that needed done? With our Frankenstein and dispirate systems, scripts do not work let alone AI automation. They won't pay to fix that.

[-]

0263111771@reddit

First, I am sorry you lost your job. I know the feeling.

Second, there are so many posts explaining the same situation and everyone responds with "it won't work" or "AI cannot do that yet".

Okay, what about when it can? At the rate AI and computing power are evolving we all know AI will get there. Not if, but when!

I think everyone is focused on the limitations of AI more than on what is going to happen when the limitations are overcome. And this is not just the tech industry that I am reading posts like this.

[-]

AliveInTheFuture@reddit

It can already do my job. Sure, it’d probably make a few errors here and there, but in an agent swarm I’m sure they’d figure it out. It would take minutes instead of days/weeks.

All these anti AI posts sound like ostriches with their heads in the sand. Either that, or they just haven’t been experimenting with SOTA tooling and models and simply use the chat interfaces for LLMs.

[-]

0263111771@reddit

Heads in the sand or not. Meta, Oracle and AwS have already laid off thousands of people and the companies themselves have stated AI adoption was the reason. Furthermore, this is the worst I have seen the IT field in 23 years. Mass unemployment and longterm unemployment at that. AI exists for one reason and only one reason, and every company is going to use it no matter if its ready or not.

[-]

BattlePants43@reddit

Hallucinations are the main issue, and we have no reason to think hallucinations will stop. It's better hidden, but it will always be there. Human in the loop is a minimum requirement for responsible AI use.

[-]

AliveInTheFuture@reddit

Hallucinations are pretty rare now.

[-]

0263111771@reddit

I agree, there will be a human in the loop. But much fewer people than jobs taken

[-]

rc_ym@reddit

If you have uptime expectations that match Claudes performance... It can probably work for technical work.

And the agentic loop in CC and Codex does really good non-coding work. I literally have a team right now that could be replaced with AI, and the product would be better. I use it to check their work. 80% of the time it's better. I'm sitting on it not to start a trend. 🤷

[-]

synthetix808@reddit (OP)

thx. [virtual fistbump]

[-]

Secret_Account07@reddit

I guarantee the execs expectations of what AI will do for them vs what it actually will do are vastly different.

AI can do a lot of technical part of job but have it go through a change request or resolve an outage at 1AM and communicate the issue/resolution efficiently to mgmt.

Tell me when that happens lol

[-]

AliveInTheFuture@reddit

It can do that already.

[-]

No_Investigator3369@reddit

With more devops and SRE eliminations lately, is there a repatriation push happening at the moment?

[-]

Skyhound555@reddit

Yes.

The reality is that DevOps is dying. Companies are regretting paying six figures for newbies who only have a AWS Solutions Architect cert. These engineers ended up silently costing their organizations millions of cloud costs with not much to show for it.

On-prem is attractive because costs are static and easily manageable. 100% uptime is also much less important now than it ever has been. Companies don't actually lose money when a site goes down, customers will usually just try again.

Also, one big point: AI will never be able to replace rack and stack skills. Low voltage cabling and server rack building is an inheritly human skill.

My company never went full cloud and that gamble paid off hugely in the end. To this day, hosting on-prem is still cheaper after we do the actual math. Yes, even with these RAM costs.

[-]

No_Investigator3369@reddit

For years I worked in organizations whose cloud seemed very fragile and did not seem to scale up or down. To me that seemed then an on-prem data center in terms of it being a liability and a line item on the budget sheet. I never went cloud and stayed very strong in sdn dc's And already have the background for the cowboy CLI vxlan stuff If someone wants to waste their time being cheap without a controller/inventory management solution. I don't think I'll take it as personal In the future and just realize it's just people unqualified for their jobs making unqualified decisions.

[-]

RumRogerz@reddit

They will eat these words. AI sucks at infra. Sorry this happened to you bruv

[-]

synthetix808@reddit (OP)

Nice one bruv. Thx

[-]

wise0wl@reddit

Our teams limitations are not on code throughput but on ensuring reliable architecture, thorough testing, and risk analysis and planning within the context of our existing systems and future plans. A human has to check these things, verify, and then execute.

In theory AI could just talk to the executive and the PM and work out the technical architecture and implementation. It’s possible. AI is not proactive yet though. It doesn’t go looking for optimization and fixes. It doesn’t have insight into deeper context that’s stored in the depth of experience a human has. If a job can be done by a contractor it can be done by an LLM, but the human / machine interface still requires a deeply technical PM, and those are exceedingly rare.

My team of four oversees four distinctly different business units, and the context there runs very deep and is often undocumented or unstructured. LLMs save us a ton of time and we are getting more done than ever before, but we still need people guiding it, placing and checking guardrails.

We won’t be replacing humans in our Platform, DevOps or SRE departments anytime soon.

[-]

Pollo_Caliente@reddit

Up vote for the Ghost in the Shell quote. Such an amazing show, still.

[-]

Chaucer85@reddit

I still remember the conversation where one of the Tachikomas, in its upbeat cartoonish voice, proceeds to lay an entire university-level philosophy lecture on Batou, and it cuts to him with this open-mouthed dumbstruck look on his face. A tank with a twee personality is talking to me about the nature of a soul. What is my life?

[-]

Mindestiny@reddit

The Tachikoma Existential Philosophy episode of SAC was one that people either loved or hated, but it sure as shit worked for the subject matter.

Can't give them the fancy oil or they might develop the concept of individuality!

[-]

Chaucer85@reddit

I was gonna say, I love that utilizing non-standard oil, leading to possibly contamination of their sensitive processing units, is what likely led to them diverging into unique personalities and arguably sapience.

[-]

TheBestMePlausible@reddit

Those cutesy sentient spider murder robots were something else!

[-]

Humble-Plankton2217@reddit

I'm so sorry this happened to you.

15 years into my IT career working for the same company, I was let go unexpectedly. There was a CEO turnover and the new CEO culled every on-site IT person in North America to save money.

The next IT job I found paid me 20% more and had a much better environment. I hope the same happens for you, that this is a forced change to something better. I stayed with 1 company far too long and found out their measly 2-3% raises each year left me way behind on appropriate payscales.

Stay strong! Wishing all the best for you.

[-]

synthetix808@reddit (OP)

Thx. Bigger and better things await

[-]

Audience-Electrical@reddit

Just got fired today too! Good luck, brother.

[-]

synthetix808@reddit (OP)

Thx yo. Hang in there. We got this.

[-]

agingnerds@reddit

I just dont see a world at this point with the AI we currently have replacing people. I was working on a cql query in confluence and the suggestion was just wrong. I found the article it used and found the solution there. Ai is interesting, but waaay too early to replace a person. I think its a solid tool to assist knowledged people, but its not ready to replace.

[-]

fartiestpoopfart@reddit

i work at a very large global corporation where there has been a huge push from the top to use AI as much as possible and i can't help but feel like we're just training it to eventually take our jobs.

fortunately for me my environment is a fucking mess and i don't think we're anywhere near an AI being able to replace my team, but the huge AI push is still weird and i don't like it.

[-]