I have been trying to come off of google and cloud by building — quite slowly — my own nas server which has 2 nodes in two geographic regions where I am building certain services like cloud storage and backup, webhosting etc. But I think there are a few key things that need to be community driven to really get rid of this duoply.
0. A privacy first approach would be something like this:
`You+App --Read/Write-> f_private(your_data) <--Write only- 3p` and App cannot communicate your data to 3p or google/apple.
Think of Yelp/Google Maps but with no _read_ permissions on location, functions can be run in a private middleware e.g. what's near an anonymous location or ads based on anonymous data. You can wipe your data from one button click and start again for EVERYTHING, no data is ever stored on a 3p server. Bonus: No more stupid horrible permission fiascos for app development that are just plain creepy.
1. An opensource data effort that can support (0) with critical infra e.g. precise positioning, anonymous or privacy preserving functions that don't reveal their data or processes to 3p.
Here is my favourite open source effort: Precise Location Positioning. A high recall, opensource, 3D building and sattelite-shadow Data-Infra effort[3]. This world class dataset on shadows and sattelites are a must. Most geo-location positioning tied to Radio signals is just a bandaid and fraught with privacy issues — thought there are heroic privacy first efforts in this direction[1][2] which though amazing will be playing catch-up with google already deploying [3].
It means any 3rd party even the app provider cannot read your data or the output of the function run. They can provide some data/resources like say map tiles, PoI data and a function to run.
All coding agents are geared towards optimizing one metric, more or less, getting people to put out more tokens — or $$$.
If these agents moved towards a policy where $$$ were charged for project completion + lower ongoing code maintenance cost, moving large projects forward, _somewhat_ similar to how IT consultants charge, this would be a much better world.
Right now we have chaos monkey called AI and the poor human is doing all the cleanup. Not to mention an effing manager telling me you now "have" AI push 50 Features instead of 5 in this cycle.
They are not optimized to waste tokens. That is absolutely ridiculous. All of the LLM providers have been struggling from day one to meet demand. They are not trying to provide outputs that create more demand.
In fact, for example, Opus 4.5 does seem to use fewer tokens to solve programming problems.
If you don't like cleaning up the agent output, don't use it?
We’d close one of the few remaining social elevators, displace higher educated people by the millions and accumulate even more wealth at the top of the chain.
If LLMs manage similar results to engineers and everyone gets free unlimited engineering, we’re in for the mother of all crashes.
On the other hand, if LLMs don’t succeed we’re in for a bubble bust.
As compared to now. Yes. The whole idea is that if you align AI to human goals of meeting project implementation + maintenance only then can it actually do something worthwhile. Instead now its just a bunch of of middle managers yelling you to do more and laying off people "because you have AI".
If projects getting done a lot of actual wealth could be actually generated because lay people could implement things that go beyond the realm of toy projects.
You think that you will be ALLOWED to continue to use AI for free once it can create a LOT of wealth? Or will you have to pay royalties?
The rich CEOs don't want MORE competition - they want LESS competition for being rich. I'm sure they'll find a way to add a "any vibe-coded business owes us 25% royalties" clause any day now, once the first big idea makes some $$. If that ever happens. They're NOT trying to liberate "lay people" to allow them to get rich using their tech, and they won't stand for it.
This. This is what I find hilarious that even smart HN folks seem unable to understand. Transformers tech products are a service offered by private companies who are under no obligation to serve it to you indefinitely. At any given point, they are free to end public access. And you better believe that they will do so if it is in their interest. inb4 open source models, those models are also hosted on the servers of private companies who are also under no obligation to maintain public access indefinitely. And even if you were smart enough to download one in advance, cloud services providers can stop providing access for transformers and you can rest assure that your machine won't be powerful enough to run it. Plus, NVIDIA and co can just keep their GPUS to themselves and only offer subpar versions to customers.
An individual will never win a fight against a corporate entity. And certainly not one in possession of a near AGI system.
Thats borderline aluminum hat conspiracy theory. Corporations arent a monolith, you think amazon is ever going to stop you from renting machines so that you cant run your AI models instead of buying from OpenAI? They have no horse in that race.
> you think amazon is ever going to stop you from renting machines so that you cant run your AI models instead of buying from OpenAI
We are talking about a future with near AGI systems. In such a future, people like you or me have no money to pay those services with because we are all unemployed and starving. And amazon has much bigger ambitions than just resting cloud compute to you. The economy as we know it doesn't really exist in that scenario and neither do the incentives and constraints that exist in our current economies.
People talking about intelligent systems a lot without considering the profound changes it would cause to everything.
There is no future where near AGI and traditional economies coexist. Near AGI is essentially a type of swan
We are not ready for social media. And we are definitely not ready for transformers let alone some sort of sub-AGI that is still powerful enough to complete most projects. Economies would fall quicker than the stock market on that fateful black monday. Our economies still operate on the assumption that only humans can do most of the work that delivers value. Remove that assumption, and you have nearly zero operating costs but also nearly zero revenue for virtually every single company operating mostly in the knowledge sector.
>If projects getting done a lot of actual wealth could be actually generated because lay people could implement things that go beyond the realm of toy projects.
Suppose LLMs create projects in the way you propose (and they don’t rug pull, which would already be rare).
Why do you think that would generate wealth for laymen? Look at music or literature, now everyone can be on Spotify or Amazon.
The result has been an absolute destruction of the wealth that reaches any author, who are buried in slop. The few that survive do so by putting 50 times more dedication into marketing than they do to the craft, any author is full time placing their content in social networks or paying to collab with artists just to be seen.
This is not an improvement for anyone. Professionals no longer make a living, laypeople have a skill that’s now useless due to offer and demand, and the sea of content favors those already positioned to create visibility - the already rich.
That would be true in a monopolistic market. But these frontier models are all competing against each other. The incentive to 'just work and get shit done fast' is there as they each try to gain market share.
I've spent an inordinate time tuning my home nas to what I want to do. e.g. NPM reverse proxy with all kinds of docker apps and configuring it securely made me appreciate how tough cyber security is. My latest venture was to create several docker images for whatever environment I want: VsCode Remote SSH server, a docker image with a custom LLM, sync thing for obsidian notes across all devices
My goal is to de Google with home nas hosting a bunch of services. I want to take all my Dropbox photos and recreate what Google memories does using off the shelf AI tools on my nas. Then email and finally maps with great PoI data.
Probably not all of what I did is a differentiator that put my learning beyond mediocrity. But a good yard stick is if what I built is useful to someone and even better if one would pay me for something.
I am not sure it's that simple. Thing is many costs are due to information assymetry: you not knowing something that they do.
For example them counting on you to big hire a lawyer for collecting medical debt or mortgage debt your spouse or parent owes. As a general rule you aren't responsible for it. There are exceptions. e.g. Filial laws(children responsible for parent's debts) exist in many states, but are difficult to invoke. Community property laws https://www.irs.gov/publications/p555#en_US_202502_publink10... in 9 states that can link your income to your partner, when the state you were domiciled in with your partner in a home/condo you bought together.
So in general, adverserial use of AI cannot bring claims "back to pre-AI" levels. Much more likely is the fact is reduced debt collection activity and illegal billing will reduce to a new baseline.
A pharmacist once told me that big insurance companies have call centers out in other countries whose job it is to call everyone who has medication approved and any one with misleading line of questioning:"Did you get or approve medication worth $$$$$$..?".
Calling 100's of people Ofc the find one poor guy never heard of such a sum denies this kind of line of questioning. Then the insurance company uses this to deny all claims made by the pharmacy for ALL their patients for that given drug/medication.
The pharmacist told me the mountain of documentary evidence they have to collect to rebut these denials is very large. Once a customer at their pharmacy said he did not want to sign off on a paper that he got a medication, the pharmacist got the customer's ok though to video record his consent, just so he does not have to deal with this mess.
He also mentioned to me that a pharmacist should NEVER pay any kind of reimbursement to an insurance company on a claim that was denied cause that somehow legally can let the insurance company deny future claims. Not entirely sure what exact legal procedure allows them to do that.
Does someone here understand how exactly to fight Facility Fees — outside of indiana or a state where its outlawed — which is what the author mentioned most of their fees were?
Could one when signing admission forms accidentally agree to paying them without fully understanding it?
After one gets the the bill can one simply get an itemized breakdown, spot these fees and negotiate them down?
The author mentioned in a reply in threads that most of the fees was Facility Fees. That might be just wrapped up in a code for cardiology sure, but its just profit chasing hospitals because that practice can wildly inflate the cost and billing and that can be fought.
What's crazier is that for 80k you can get a jet to fly you anywhere in the world and for far lesser than that get world class treatment. What's even more bonkers is that the private and govt insurance companies, and hospitals have negotiated those rates and there is a market to fly people to other countries that is just sitting there and no one is really exploiting it.
Honestly, that's not a bad idea for a start up. Maybe a marketplace where people can see what things cost in different places and book a surgery directly.
