Fuck AI

I think an even bigger problem here is - what are you bringing to the market, and why should you be making any money? Your AI provider has perfect knowledge of how your "organization" works, but they are far better equipped to optimize it, while you're just using their API as a black box.

It depends, but it's anyway worse for employees as for the company.

Say you are a company that hunts and catches whales. You can sell the whale remains at a profit.

Whale numbers are finite and declining.

Now, you can catch and kill only a few whales per year, stabilizing the existence of natural, um, resources, to make your business sustainable.

Or, you can catch as many as you can get, as quickly as possible, destroying the resources your business runs on. (And killing a species of magnificient social living beings).

Which path will a capitalist company chose?

The idea AI companies could have so much leverage is why investors value them so highly

I don't see it though, for the simple reason that open source models are only months behind

I expect to see commodity pricing in the long term

The goal is the eliminate services and products that regular people need. They don't want to sell things to us. They want to sell things to each other.

Even if they eliminated all the employees or most of them, the problem then becomes, all the infrastructure can't be done by AI or upkept by AI.

People are needed for that. Even if you only use a skeleton crew they need things. And they know the infrastructure too. So that means there's no way to not provide for at least some people because that work is still necessary.

Which means there's still consumers and they require services and products. Necessities at the very least.

It's turtles all the way down. Even if it worked the way it's marketed you create a new problem. People who know how to upkeep these systems who either strike, or mess with the system intentionally because they aren't having their needs met.

Even ignoring the rising cost bait and switch it's a terrible idea.

It's a race to the bottom. Only the ones who do it early can get rich in the process, while the damage can still be diffused. The more do it, the more damage there is, the faster it gets worse.

Not thinking big or evil enough.

You're a company with 1000 employees. A company like anthropic is offering to replace each employee with some sort of agent. You accept and let them train models on your processes. They launch a competitor company and put you out of business. End of story.

You go bankrupt, AI company buys you and all your competitors. Other AI companies do the same. Each product includes AI costs like other raw materials. All costs passed down to consumers (which have no money to buy shit coz they were replaced by AI). AI companies just trade between each other in walled gardens.

Business is fine. People are screwed.

Some other scenarios to consider:

• AI company decides to up the profit by switching your agents to a cheaper, worse model behind the scenes while keeping your costs the same
• AI company deciding that your business is plenty lucrative and now that you've onboarded their agents and they know your day to day, what do they need you for? You're just a shell around their agents. So they cancel the contract, create their own competing business and poach all your customers.

the interesting bit is, 90%+ of ai revenues come from a few big companies, not regular Joes. And i think ai is a way for some companies to extract every drop of value from other companies, now that people are squeezed dry. I guess companies turning on each other, which might be fun to watch, though I doubt we will just be on the sidelines. It's the logical next step in capitalism, if you run out of people to cheat and rent seek from, there's plenty of easy corporate victims. Sure at first it's, smaller mom and pop stuff (I mean, that's been going on for a while now), but no reason not to do it for bigger companies, might even have less overhead and easier pickings. And ai is perfect for that.

Those are great points. The equation is more complicated though. Agents can work 7x24x365, never needing vacations or time off. Never getting sick. You can scale up or scale down instantaneously without having to fire anyone or hire recruiters or temp agencies. They never sue you for sexual harassment. They never complain. You don't need office space for them, or parking, or coffee. You don't have to take mamby-pamby management classes to learn how to boost you DEI or "EQ". These things are a real capitalist wet dream. Especially when there are interchangeable vendors with no clear differentiators. Even more so if you don't care about "customers" either. If you sell to the government or other businesses or even just to other divisions in your own company, the customer is irrelevant. Only the bottom line matters.

This is literally their business plan you are describing.

1. Hype AI as the literal second coming of Jesus
2. buy all the electronic components on earth. Make it hard to impossible for most people to buy computer hardware for any purpose unless they’re already rich.
3. convince enterprise that humans are no longer necessary and that you can just use robots to do all that work and split the same profit between far fewer of you. Middle management and up cum immediately.
4. offer your cloud services to everyone who can still afford anything. You watch and profile their every move. No one dares to talk shit about you or you debank them, completely cut off from all modern conveniences. They rent this privilege from you.
5. increase prices until you have all the money.‘you now rule the world. The plebs all starve.

Of course it’s psychotic shit but that doesn’t mean they don’t believe it

Also, you are destroying the market: these 500 people will not afford to buy your products anymore and AI will not replace them as customers. Multiply that for every company in the planet: who is going to buy anything anymore?

I've tried to make the same argument understood with cloud services and vendor lock-in generally. It's crazy how companies find a few percent pay increase for employees onerous, but are perfectly willing to give the life-blood of their services to a vendor who can arbitrarily raise prices down the road. Lots of companies are seeing this already with Broadcom and VMWare

Wouldn’t replacing employees with AI actually be catastrophic for businesses?

Yup.

For one thing, LLMs can't replace employees. Not really. Some who have tried (like Uber) have regretted it. (Others have fallen hopelessly into AI psychosis.)

I think there have been several companies that have decided to lay off a bunch of people and used the lie "we replaced them with AI" to spin it in a way that makes their stockholders happy. In fact that might be the majority of companies who are reporting that they're replacing employees with AI.

But there are definitely other companies that have simply drunk the Kool-Aid as well.

Now anthropic comes to me and says they’re going to increase their prices to $100k/agent. What now?

Price increases is definitely gonna happen, actually. And not just because all the "AI" companies are greedy bastards. All the LLM offerings out there are running at massive losses. They're trying to get people/clients hooked so they can way increase prices. And they'd have to increase prices a lot to make it actually profitable for them.

Each model has its own strengths...

That's putting it a little generously for my taste, but yeah.

There is a back-channel that's not mentioned here.

The executives and directors of these companies all play at the same golf clubs. They know each other.

So, when this happens, they just have a chat at the 19th and the prices are negotiable.

The smaller businesses are the ones that get screwed. Not to mention the ex-employees.

It's still as stupid as it sounds.

This is about timelines. In the short-term it's a win, in the medium-term it's a threat, in the long-term it's potentially catastrophic. They need money now.

This sort of scenario assumes that these companies will always be in a stronger position. I don't think that's a given. The companies that plundered all data everywhere, torrented what was available, and reluctantly bought what couldn't be stolen to train their models are being passed on the right by companies that just look to steal or copy their stuff (e.g. China is inferencing the hell out of them). The hostility towards data centers also forces more development away from centralised servers onto local environments. At the same time, stuff is also being released as open source. I think there is a big chance that the big companies and their models today will be mentioned in the first chapter on the history of this technology but will fall by the wayside rather quickly. Their market position is underpinned by a belief that in the future they will be raking in the dough on subscription fees once they get the dirty work of being started here financed, no matter how. That's a mighty big bet. Not only is a bubble economy a problem, there is no guarantee to protect their "IP." It's really like investing in very expensive tulip bulbs.

Also, we don't fully understand this technology. A lot of the experts don't either. Anybody who says they're pretty sure they know where this road will take us in the next five years has ulterior motives.

I honestly think the best application for ai would be management

But quarterly line must go up! 📈

Your scenario is a good point. There's also another one - that you invest in this company to carry your load and something happens to them, so they don't raise prices, they stop providing the services. Now, maybe you can quickly switch to someone else, maybe that costs a lot to do and time as well. Maybe it affects all AI companies, and you're screwed unless you can replace them with humans, probably asking even more than before.

Also another one - the original agents worked well. Then they try to go full Microsoft and do an upgrade which for whatever reason breaks things. Maybe the agents are worse for what you use them for, or maybe it's just an incompatibility. Either way, damage control again, where if you had stuck with at least some or most of the humans you could weather.

Eggs in one basket. Humans bring diversity, flexibility, and adaptability, even if they cost more.

Who cares about future issues. I've got my golden parachute until then!

Yes.

It's probably going to be closer to 500$ a month. Going from 100k to 6k is a lot more enticing, even with the eventual price hikes. It's never going to start so high, they aren't stupid. Their end goal is to have a blanket AI tax on most businesses and capture a significant part of the economy.

The thing is though, if the company can just give them the finger, build a computer for 5k and run an open model on that, then Anthropic and friends lose their whole business model.

Hence the media campaign, buying supplies like raw wafers, boosting the price of entry with the strengthening of copyright laws and stoking fear to have a government mandated pause, etc.

It's a shame but the anti-ai movement usually aligns itself with what big AI wants (pro copyright, pro regulation, fear mongering).