Let’s say for a moment that AI really is as capable as AI bros want you to believe it is. Having used some AI models at my work (network engineer), I’m not so convinced, but I’ll put that aside for a moment.
The pinch of the AI vise is going to be the cost. All the AI companies are running at loss to try and capture the market, and they’ll need to charge a lot more for access to AI once a winner is determined.
Are you willing to pay 10x what you pay right now for access to an AI agent? Because that’s looking like what they’ll have to do in order to turn these companies profitable.
Are you willing to pay 10x what you pay right now for access
That’s really the crux of the matter. Regardless of if it’s fit for purpose to replace paid labour, is it actually financially viable in any way to replace humans with an AI model that works but costs several times more than a human work force while not being able to be held legally accountable.
That only makes sense if money is made on the output. What you’re basically saying is X Toy Company will make more money if they make more toys. What you’re forgetting is that there has to be a demand for those toys. So X Toy Company can’t replace workers with AI unless there’s more demand for the toys than there is currently able to be fulfilled and there’s no evidence for that across all sectors of the economy.
I pay zero, not even electricity because it’s being generated in this building, for my agent. It’s running entirely on my computer and it’s decent enough. The LLM subscription companies are eventually going to fail because everyone and their dog is going to be able to run their own models.
Our company built an “AI” system that literally only searches our other shitty system and gives a usable interface. It is not AI and im sick of every database interface being called AI.
LLM at least makes sense from a nomenclature perspective.
10 years ago it was “smart thermostats” “smart x” just because they could connect to the internet.
This is the same gilded bullshit these companies always spin.
They’re playing all kinds of games with the numbers right now. They’re trying to IPO so hiding costs under other headings outside the core business. Don’t trust what they openly state. It’s far worse.
The data is there, but it’s not in what those companies say.
You can’t argue from the premise of conspiracy. Are they exaggerating? Maybe? But you absolutely cannot state with confidence that you know a huge secret hidden from all Wall Street analysts.
$20B loss on $50B revenue are the numbers that Wall Street thinks is correct. That’s the data you have to go off of without unsupported conspiracies.
In theory the “reasoning” part doesn’t need a knowledge update.
Cost/benefit maybe the knowledge update only needs something every 5 years.
Basically I can see a world where inference is provided and consumed at cost. I can’t see one where customers are willing to cough up billions for a slight increase in the knowledge base.
I’m comparing the whole of human knowledge to 5 years of additional facts. Most useful things are known. An llm doesn’t need to know the current US president to invent a biscuit recipe.
I’ve already seen it replace people in various roles successfully. It’s getting better – will it replace everyone? No, but it will be disruptive and it seems foolish to not acknowledge what is obvious - jobs will be affected.
They’ve been shown to be a mathematical inevitability, and that means everything they ever make must be fully reviewed.
And that’s antithetical to everything I’ve known about computers in my career.
The entire reason why computers are so useful and reliable is because, until now, they always produce the same outputs given the same inputs.
One of the AI agents I’ve interacted with was in a GPS monitoring system for fleet vehicles. I was a supervisor of technicians, and I got a report that a company truck with license plate [X] ran a stop sign.
They replaced their standard search function with an AI bot, so I asked it which vehicle in the fleet had license plate [X]. And it confidently responded with a vehicle that didn’t have that license plate.
A standard search function, comparing an input to database values for vehicles wouldn’t have that problem. And that’s why I’m so sure this is a bubble and not a major technological market disruption.
Let’s say for a moment that AI really is as capable as AI bros want you to believe it is. Having used some AI models at my work (network engineer), I’m not so convinced, but I’ll put that aside for a moment.
The pinch of the AI vise is going to be the cost. All the AI companies are running at loss to try and capture the market, and they’ll need to charge a lot more for access to AI once a winner is determined.
Are you willing to pay 10x what you pay right now for access to an AI agent? Because that’s looking like what they’ll have to do in order to turn these companies profitable.
That’s really the crux of the matter. Regardless of if it’s fit for purpose to replace paid labour, is it actually financially viable in any way to replace humans with an AI model that works but costs several times more than a human work force while not being able to be held legally accountable.
If this future AI agent solves the task at a fraction of the time, it can still be financially viable at several times the cost.
That only makes sense if money is made on the output. What you’re basically saying is X Toy Company will make more money if they make more toys. What you’re forgetting is that there has to be a demand for those toys. So X Toy Company can’t replace workers with AI unless there’s more demand for the toys than there is currently able to be fulfilled and there’s no evidence for that across all sectors of the economy.
exactly, depending on the case it’s a possibility and not a guarantee, which is why I also wrote it can be, not it will be.
Riiiiight…
I pay zero, not even electricity because it’s being generated in this building, for my agent. It’s running entirely on my computer and it’s decent enough. The LLM subscription companies are eventually going to fail because everyone and their dog is going to be able to run their own models.
Not of they can’t buy computers any more.
I’m wondering if we’ll get a rise in old school tech due to high cost of ram, CPUs and gpus.
Windows will struggle if that’s the case.Linux will be fine.
even better is the open weight, free models you can tailor to specific needs, which directly undercuts these massive “ai” companies.
So if/when those hikes happen, those self hosted models are going to look a lot more appetizing too
I think you’re probably a couple of orders of magnitude too low.
Our company built an “AI” system that literally only searches our other shitty system and gives a usable interface. It is not AI and im sick of every database interface being called AI.
LLM at least makes sense from a nomenclature perspective.
10 years ago it was “smart thermostats” “smart x” just because they could connect to the internet.
This is the same gilded bullshit these companies always spin.
They lost $20B on $50B revenue. That means they only need to double the price to make up the loss and have a profit.
They’re playing all kinds of games with the numbers right now. They’re trying to IPO so hiding costs under other headings outside the core business. Don’t trust what they openly state. It’s far worse.
Maybe but you have to work with the data you have, not the data you want to see.
Imagining that the enemy is on the brink of collapse isn’t a good mind set when you are fighting something. It encourages laziness.
The data is there, but it’s not in what those companies say. It’s in the books of the companies they have relationships with.
I do agree with your second statement though.
You can’t argue from the premise of conspiracy. Are they exaggerating? Maybe? But you absolutely cannot state with confidence that you know a huge secret hidden from all Wall Street analysts.
$20B loss on $50B revenue are the numbers that Wall Street thinks is correct. That’s the data you have to go off of without unsupported conspiracies.
That’s still a ton when you consider how much professional users are spending.
I’m fairly certain these numbers are made up. Additionally, I’m betting those numbers include investment dollars, as opposed to paid revenue.
That is if you include training. Look at deepseek for an idea of inference costs. At that level AI coding is sustainable (but not worth $2tn).
Why wouldn’t you include training? Training is vital to their product.
Is it? It’s vital to the tech giants marketing and trillion dollar valuation. But I don’t think it’s vital to most of their clients anymore.
If you want it’s “knowledge” to be up to date, the training never ends.
In theory the “reasoning” part doesn’t need a knowledge update.
Cost/benefit maybe the knowledge update only needs something every 5 years.
Basically I can see a world where inference is provided and consumed at cost. I can’t see one where customers are willing to cough up billions for a slight increase in the knowledge base.
5 years ago we were mid-pandemic. That’s a big knowledge gap.
I’m comparing the whole of human knowledge to 5 years of additional facts. Most useful things are known. An llm doesn’t need to know the current US president to invent a biscuit recipe.
I’ve already seen it replace people in various roles successfully. It’s getting better – will it replace everyone? No, but it will be disruptive and it seems foolish to not acknowledge what is obvious - jobs will be affected.
You literally said “replace all human labor” though.
Haha I’m not a AI Stan. I am the only one it seems truly worried about never having a job again.
Their investor pitch is this technology replaces all human labor. That’s their framing not mine.
My honest problem with AI is the hallucinations.
They’ve been shown to be a mathematical inevitability, and that means everything they ever make must be fully reviewed.
And that’s antithetical to everything I’ve known about computers in my career.
The entire reason why computers are so useful and reliable is because, until now, they always produce the same outputs given the same inputs.
One of the AI agents I’ve interacted with was in a GPS monitoring system for fleet vehicles. I was a supervisor of technicians, and I got a report that a company truck with license plate [X] ran a stop sign.
They replaced their standard search function with an AI bot, so I asked it which vehicle in the fleet had license plate [X]. And it confidently responded with a vehicle that didn’t have that license plate.
A standard search function, comparing an input to database values for vehicles wouldn’t have that problem. And that’s why I’m so sure this is a bubble and not a major technological market disruption.
Jobs have already been affected, yes. Not “adequately” done in lieu of a human, though.
And it still takes a human to steer the things.