16,374 karma · joined December 30, 2018
Same name on libera.chat and OTFC
The economy of the future will be about yachts, supercars, hermes bags, private air travel and ozempic. Not about how you made it possible for everyone to watch a gigantic library of videos for free, or how you made cars affordable for everyone.
It is already a quite smooth experience, but there is work to make it even easier than that.
I work on Lakebase, opinions my own.
There is no inherent property to those model makers that they will capture the value of the AI boom. The model companies have a lot of competition from chinese models. Of course the US government will likely still keep some model makers alive (just alone for DOD supply chain reasons), but the chinese models will certainly compress margins.
It might as well be a chipmaker, a power company or a cobalt mining company.
Ultimately, it's up to the US government to decide if OpenAI can capture the value provided by AI: if chinese models get banned from the F500, then maybe one of america's famous duopolies can establish themselves.
But this hinges on congress mostly to pass certain laws to put OpenAI into that position. But which value does congress have? If OpenAI is owned by a bunch of Forbes 400 people like right now, then of course those will lobby the hell out of congress and ensure their stakes in OpenAI make them extremely rich.
But Norway doesn't have that access to lawmakers, they would see it as external influence by an alien power. They can certainly benefit from this trend though by being a shareholder.
To give a concrete example of the US government making life hard for foreign investors, take the T-mobile and sprint merger. T-mobile is majority owned by Deutsche Telekom AG, whose biggest shareholder are various institutions of the german federal government.
In this case, the US government just had to _approve_ the purchase, something very standard that happens all the time, but they still dragged that approval out as long as possible, even though Germany is a fellow NATO country.
> RWE U.S. Offshore secured these leases from the U.S. government with a long-term commitment to develop offshore wind capacity in American waters and invested more than $1 billion toward the leases and the development of these projects. These leases provided legal rights to develop offshore projects and represented years of planning, investment, and partnership with federal agencies.
> After careful consideration, it was determined there is no path forward to permit these projects in the U.S. for the foreseeable future. The settlement resolves RWE U.S. Offshore’s legal claims and provides $1.22 billion in settlement funds.
https://www.rwe.com/en/investor-relations/financial-calendar...
But yeah, most problems in physics, chemistry or biology require labs on top of actual hard thinking. You need to be able to design experiments in a certain way. Once you have the funding, the right tools, the right people to use those tools, then you can use LLMs to increase the speed of the calculations and so on.
There have been other discoveries though by deepmind: https://deepmind.google/blog/millions-of-new-materials-disco...
The German car industry will not die completely, as the EU can protect its market from chinese manufacturers through regulation. But it will shrink and transform. Probably also back to fossil fuels (yuck), because there the Germans still have a lead (they are trailing with EVs).
Germany's new hot industry is defense technology.
There are gazillions of candy crush remakes, but Royal Match, launched in 2021, still made it to the top in revenue. It was not thanks to some existing monopolist pushing it, but thanks to a) ginormous marketing budget and b) relentless execution and experience of making mobile games before, i.e. you know what works, what doesn't, how to extract maximum amount of revenue, etc.
https://naavik.co/digest/royal-match-finding-success-through...
https://naavik.co/digest/why-dream-games-success-is-a-challe...
Startups are taking longer and longer to reach public markets, so more and more value creation requires you to be a family office, angel, or hedge fund.
That constant low level stream of HIV viruses in your body can cause the virus to return at any moment you stop the therapy. So you have to take daily pills for the rest of your life to stop that stream from exponential growth and killing all of your immune system.
The last part is what is referred to as AIDS: your immune system is so weak that even the simplest pathogens can kill you, which they usually do once you reach AIDS stage.
There are a couple of viruses that can lay dormant in you and cause problems later on. Many of them insert themselves into the DNA of long living cells.
Without therapy, HIV will progress to AIDS in a couple of years in basically 99.9% of patients. Some very rare cases are known where HIV patients survive without getting AIDS and therapy, via broadly neutralizing antibodies. But again, less than a tenth of a percent. Everyone else would die without a therapy.
Being HIV positive isn't a death sentence, if you take your meds (and started taking them early after HIV infection) you can live for as long as someone who isn't HIV positive. In fact studies show that the effect of regularly going to a doctor and getting your blood checked actually improves your life chances. HIV positive people can do 99% of the things normal people can, although they are not allowed to donate blood (for good reason).
Having AIDS though is a major impairment and you are on your way to death at that stage (1-3 years to live). So it makes sense to distinguish the two.
Indeed it's nothing hard to learn but there is a learning curve. E.g. knowing which model has which capabilities, and figuring out how to best manage context, permissions, worktrees, etc. There isn't one "right" way to use it but there are more efficient ways and less efficient ways.
Also, even if it were a solved problem, the methodologies developed here can probably help with other viruses too in the future. This has already happened with the antivirals originally developed for HIV.
Aren't they shredding only the books still under copyright protection? How is an 18th century botanical text still under copyright?
IDK about the shredding, it's not nice, but it's more a problem with copyright law than AI companies.
Scanning books you own should be legal from a copyright point of view, and not require shredding.
Second, one should think about abandoned property provisions for copyright works published more than 50 years ago and in danger of being forgotten: once challenged, either you as the owner have to prove that the work is preserved for future generations (e.g. in various libraries around the world), or you have to authorize further copies, or you give up copyright on the work.
With software there are two things: the source code, and the binary.
Firmware usually has restrictions on modification, so it violates the four freedoms.
Open weights is not perfect, but at least you can finetune as you wish, and for example add more knowledge about the current day to those models, or give it knowledge about company internals so you don't need to put stuff into the context.
Many smaller AI labs finetune chinese models and release the result. Finetuning is very valuable!
And yes, laptop specs haven't changed much and this is partially because the need for spec changes wasn't present, but also during the last 20 years there has been tremendous pressure for efficiency in datacenters.
Despite that, dennard scaling is dead since 20 years. There are physical limits. Already now, the wear effect of electrons jumping is present, and it will only get worse as things scale towards smaller sizes.
There are some benefits to be had, e.g. one can etch models into chips directly so you can pack them more closely, and run more inference on Tensor like chips, but that gives you maybe one order of magnitude improvement in total, at most. Also, of course nobody does that when each 2-6 months a new model comes out.
* language issues. Many chinese don't speak english. Also a problem in many european countries (esp latin and slavic speaking ones), but at least the european languages are easier to learn. Compare this to Amsterdam, Goteborg, Berlin-Mitte or Kopenhagen where everyone speaks english.
* citizenship is one of the hardest to get in the world.
* I heard complaints about onboarding into the chinese app/digital ID ecosystem.
Europe is in its own set of problems and it is not in the same situation that US used to be after WW2 (only major economy not affected by bombing).
Europe's problems:
* active major war in Ukraine (lasting longer than Axis/Soviet war in WW2)
* energy supply issues (unlike US it's not energy sufficient and the places that supply it with energy are involved with wars)
* a wall of people aging away from employment and into doctor's and hospital waiting rooms (forcing less investment into research and roads/bridges/railway, more towards stabilizing pensions, healthcare)
* major pieces of the european export economy are being replaced by China (eg chinese car brands eating the lunch of european car brands).
1. the cloud moat is mostly around talent really. Try finding people who can self host the alternatives to S3 et al at the HA and the scale the businesses need. Those alternatives are usually not free either, and each product might have its creator acquired (and the product cancelled) or similar. if you're a larger business then the data lock in becomes a moat: getting your data out of the cloud is prohibitively expensive. Furthermore, large businesses have sweet discounts.
2. ms office has immense networking effects due to its formats being quasi standards in many industries. try sending an odt to a government entity. As for gsuite, it uses open formats but it's classical google fashion a large suite of software bundled together and not that expensive for what it offers.
3. Linux is not a free alternative if you're a business, you still need to pay someone to support the computers with linux on it, and operating systems have the strongest network effects ever. Linux also has no stable ABI so one can't easily deploy third party software for it.
What's the LLM moat? Codex is OSS and Claude has gazillions of alternatives. Cursor is a nice app but it's a bunch of patches on top of vscode, a team of 5 people can vibecode it in 6 months.
If Anthropic can block distillations somehow (which are fair game imo given that Anthropic et al did the same with the written works of mankind), then they might stop or slow down the chinese from catching up.
Chinese also have like 40% of the AI researchers of the world, plus they have access to a lot of cheap labour for writing training data. I'm sure an hour of training data creation from one of China's 162 million university educated people is much cheaper than an hour of work from one of US's 97 million. Probably still cheaper than someone from the grand area.
China is behind in AI chips/GPUs but they are catching up. One thing where they have a hard dependence on outside is their energy imports: they have to import a lot of stuff from third party countries. The US on the other hand is energy self sufficient.
It gets abstracted away for richer countries as they can outbid the poorer countries for food. In developed economies, most of a particular piece of food's price is not the costs that go to the farmer, but costs that come later in the process, so that cost increasing is also felt less.
Heatwaves on the other hand affect the western countries directly.
Windows media player probably sees very little usage nowadays and probably even less for HEVC, when most content playback happens via streaming and browsers today.
As for the RAM increase, well that's probably a consequence of the general trend of doing frontend engineering via JS/TS instead of using OS native frontend APIs. The advantages are more on the development side of those apps, i.e. you can hire JS UI devs way more easily, and probably LLMs know way better how to deal with a react app than an UML one.
[1]: https://arstechnica.com/gadgets/2026/04/lawsuits-licensing-a...
So one needs to figure out a delivery method that is efficient enough, and that doesn't elicit an immune response. But I guess one can analyze the cancer in the lab and figure out which receptors it expresses, and then bind to those? We could have a toolkit of different delivery methods, tailored for each patient's cancer.
Software engineering was a nice target because inputs and outputs are just data and you don't need to figure out robotics. But idk, 3 years ago it seemed illusory (at least for me) that LLMs could take over software engineering, but now here we are. They are still not 100% there yet (software engineers still have jobs), but we are getting ever closer.
Companies are in the process of figuring out robotics, and even if it's not figured out, then we might introduce a gig-ified blue collar economy where an unskilled, underpaid gig worker implements instructions by AI. Plus a lot of blue collar work already today involves robots (cranes, excavators, trucks, etc).