Corporate America seems to be pricing in a massive productivity boom where headcount goes to zero, but the reality on the ground is completely different. The narrative that AI is about to wipe out the white-collar workforce is mostly hype being pushed by the companies selling the chips and the software subscriptions.
Look at actual enterprise deployment, and the story completely changes. These models are incredible at summarizing text or generating boilerplate code, but they are still fundamentally unreliable for mission-critical tasks. If an employee messes up a financial model or a legal contract, there is accountability and corporate insurance. If an AI hallucinates a regulatory filing or breaches client data, the company is still on the hook for billions. Legal departments are not going to sign off on autonomous agents handling real money without a human babysitting the output.
And then there is the execution gap. The hardest part of most corporate jobs isn't just generating the initial output; it's navigating the friction. It's getting approvals across departments, reading the room during a client pitch, or pushing back on a terrible idea from middle management. AI solves the easiest 80% of a task, but that last 20% still requires a human. Since the human still has to be there to finish the job and take the risk, the headcount doesn't actually disappear.
Underestimating the basic economics of compute is another massive blind spot here. Running massive inference models at scale for every mundane corporate task is incredibly expensive and energy-intensive. For a lot of operations, paying a human salary is still cheaper, far more flexible, and carries less technical debt than building, training, and maintaining a bespoke AI architecture.
Don't get me wrong, the productivity gains are definitely real. But they are just going to make current workers faster, not replace them entirely. The market pricing in a rapid, seamless transition to autonomous enterprise operations is going to get a harsh reality check.
Eventually, earnings calls are going to have to admit how hard the actual integration process really is. When Wall Street realizes that deploying this tech at an enterprise scale takes years of grunt work and doesn't magically zero out payroll, the valuations on these infrastructure plays are going to violently correct.
Unpopular opinion: The market is massively overestimating AI's ability to replace human jobs anytime soon
byu/weightedslanket ininvesting
Posted by weightedslanket
29 Comments
Corporate culture is using the narrative to squeeze productivity out of white collar…..
Like earthquakes, everyone knows it’ll happen but we just don’t know the timing.
It would be crazy to say AI doesn’t improve productivity even with today’s models. I’m not going to claim it’ll reduce head count to zero, but it does help real people do their job faster.
Is that an unpopular opinion ?
Today you are correct, but markets are forward looking. I have no idea what this technology will be in 6 months, not to mention 3 years or 5 years.
I think things are way overhyped and the market is overly euphoric, but I am also afraid enough of AI as it currently exists to wonder if my job exists in 2029. And that’s what the risk-takers are betting on.
An 80% reduction…even 20% across pretty much every sector is catastrophic economically. No one thinks AI is ready to replace an entire workforce at this point, or even the next couple years, but large reductions will be a problem and the lack of significant competition(diversification) in the majority of sectors will be a problem for workers. And productivity gains haven’t led to higher pay in half a century now so, so that ain’t gonna save anyone either.
You really think this is unpopular?
At best right now you can maybe hire one less person if your remaining staff are proficient with AI
Do you have specific companies you think are being pumped up by this that specifically would rely on AI replacing humans?
My perception is AI is overhyped, and there’s very risky things going on with scale up/capital expenditure to build out data centers/etc assuming revenue scaling/timing. But, I think the companies making money off the hardware/build outs are making out on the narrative and others investing.
The other thing is AI doesn’t need to replace humans to have value. I think the companies that are looking at how to use it to amplify productivity/capability are the smart ones.
This is more or less right from my viewpoint inside a big organization.
I think at most this is what happens:
* Some companies use this as cover for layoffs they wanted to do anyway (happening now)
* Hiring is down to flat for a few years as productivity is absorbed (maybe happening now—needs more data)
* Entry level roles are screwed and the next generation is going to have to find different ways to get on the ladder, differentiate themselves and add value (definitely happening now as shown in unemployment figures for specific majors)
Anyone who says AI can fully replace a person has no idea what it takes to actually get things done within an organization. It’s not just write a clean automated process and generate some code.
If you look at what integration and use cases were even 6 months ago compared to now – it’s evolving very quickly and companies are putting people’s jobs on the line to accelerate the adoption. It’s not ready now but I think markets are reacting to the speed in which it’s evolving. Data center land approval and power infrastructure seem like bigger wildcards.
That’s literally the gamble isn’t it? All this tech, spending, building….If the bots are running marathons in China what will they be doing next year?
AI has been an easy scapegoat to justify economic related layoffs
Unpopular opinion maybe, but AI doesn’t need to be good enough to replace the average worker to provide a lot value. It just needs to be good enough to replace the average outsourced worker in India and Philippines who barely care about their job
Only on reddit will someone have the bravery to take an unpopular opinion that 90% of the community agrees with
Bravo
This is boilerplate analysis 😵💫Long time ago it was calculated that avg. white collar worker truly works 2-3 hrs a day. 50% work really 0. AI is not excuse to replace but final straw to make it happen. Blue collar workers produce on avg. 6 hrs a day. So they still hold the valaue
You are overthinking this. Corporate America is bloated, has many people who’s job is either to do nothing, stare at the wall during meetings, or just copy data from point a to b
They didn’t even need AI to get rid of that last group, good automation would have made a lot of people redundant. But automation products like dataiku are so bad you end up needing people full time to understand it an fix it when it breaks
So first, this is just another opportunity to get rid of the dead weight, people are like “why fire 20% when profits and revenue are up” but **that’s the point**, if I can make more revenue next year with half my staff why the hell wouldn’t I?
Then you have ai helping replace the data copiers.
In general I’d say corporate America can survive fine with 33% of white collar workers gone. And that sucks for people who are used to hanging out in the office talking about fantasy football, but that’s just reality
A 5% reduction in employment would already be catastrophic. That would mean a 11 % unemployment rate in most countries.
People thought the cotton gin would end slavery yet all it did was push demand for physical labor even higher
Every productivity gain is accompanied by mass unemployment prophesies, and eventually people just find more productive things to do than easily automated tasks and life goes on
People overestimate affects of technological advancements in the short term and underestimate them in the long term. The difference here being the speed at which it’s happening
It doesn’t even matter if AI costs more than humans. They don’t want the liability.
You haven’t used “AI” enough of you think it’s going to replace jobs… That is, high quality jobs. Low quality ones no one cares about anyways.
For now it is. But not sure about the future. Part of the challenge for full AI adoption is cost. It is proving, especially Claude Enterprise, to be wildly expensive. While it may be able to replace 10 people, the cost can be around the equivalent of 12 people, so while there’s major value in speed that it provides, it isnt yet meeting what we all thought it would.
But let’s not forget that AI is still in its infancy.
LLMs today can’t replace judgement but this doesn’t mean they won’t replace a huge chunk of white collar jobs. I work in tech. We are the canary in the coal mine. I see it every day how we are rewiring our company for AI. Tech is the earlier adopter, but it will spread to other companies over time. If we ever solve AGI, I think we head towards some dystopian future. Let’s hope we are decades away from that.
Depends on the job
Programmer and customer service are highly as risk
Is the “price” reflecting humans being replaced and a boom in revenue due to lower payroll or is it instead pricing in the increase in efficiency and production?
All this human replacement by ai is bullshit. Government relies heavily on income from personal income tax filings. Good bye jobs means governments go insolvent. Corporate tax is tiny compared to what individuals pay for. Corporations are not going to pay both .
You know that person on your team who you don’t really know does what. But is kept on to do some rote task because that persons skill set is kind of lacking?
Yeah, that’s what AI is replacing and is like 50% of the workforce.
I think it will reduce the need for entry level jobs.
What are the highest paying jobs of last 10 years? Software engineering/Data Science?
If AI brings even 30-40% efficiency to this domain, it is going to be a big gain for shareholders. It’s another thing thst cost of AI is going to offset any such efficiency gain. But efforts are on to reduce the cost in future. So all the investment is not without hype. That’s the unpopular opinion…