it’s so amazing, the absolute brain rot it takes to think that a LLM is a better way to operate a vending machine than simple if-then logic. “If the value of money inserted is equal to the price, then dispense the item”.
Like, why? What is even the point? It doesn’t need to negotiate the price, it doesn’t need have a conversation about your day, the vending machine just needs to dispense something when payed the right amount.
they’re gradually reducing the amount of vehicles sold while increasing the margin per vehicle. They’re approaching the limit of 1 vehicle sold at infinite margin.
Because it’s not about the companies being profitable, it’s not about making products people want to use or pay for.
It’s about riding the hype cycle to maximize share price. Because the people making decisions are not payed based on the success of the company, but on the success of the share price and market cap.
Their current business model is not “not very profitable” it is deeply unprofitable.
They aren’t just loosing money on free users, they’re loosing money on payed users. Publicly traded for profit companies are legally obligated to provide accurate reports of the nature and source of their revenue. As of this summer (Second quarter), OpenAI was paying roughly twice as much servicing the demand of paying chat GPT users to publicly traded companies, like Microsoft, as OpenAI claimed to make from subscriptions to chat GPT.
And that’s not even counting the costs to train new models, spending with private companies, or their spending on building data centers with Coreweave or Oracle.
I highly doubt that adding advertising revenue will close that gap, especially since paying users might cancel their subscriptions if they start getting ads.
Sounds interesting, but, I do worry, if such a system were to get any sort of significant adoption, it would create a financial incentives for projects to do questionable things.
Like, even the best intentioned dev would have a very strong incentive to intentionally make their software run in the background in a way that made it look like it was being used. And if a lot of projects did that, then, suddenly there is a bunch of always on stuff eating up system resources.
There is also potential complications around one project pressuring or paying off others to do stuff that gets them more run time. Like say, pressuring a distro or desktop to include their project as a default that turns on when ever the system starts. Or simply include their project as the default even if it’s not well suited to the task.
The incentives created by the system for devs and projects would need to be considered in aggregate, like what down steam outcomes could be created for the entire software ecosystem.
KDE is avalible for most distros. It being just a desktop environment. It’s well supported on Fedora, openSUSE, Debian and Arch. As well as many of the various distros based on those. Ubuntu, a Debian derivative, and fedora both have a version that installs with KDE out of the box, and the arch install script has it as one of the main options. You could also install it on mint, but, like, half the point of mint is the cinnamon desktop.
If you’re interested in customizability, are willing to read some wiki pages, and never want to wait for support for some new feature, arch is great.
If you want a system that’s incredibly stable, will run on basically any computer made after 1995, and is generally just very reliable. Debian can’t be beat.
Fedora and Ubuntu are both fairly easy to use, new versions are released fairly often. If you don’t want to think much about it, they’re good options.
As for game compatibility, most will work without any effort, some stuff will need a bit of puttzing with settings. The only situations where you may need a VM or duel boot would be certain competitive multiplayer games that specifically use kernel level anti cheat. If you play one of those, check it on ProtonDB . Notionally Proton DB is for the steam deck and steam games run through proton, but generally what’s on there also applies to any other game run through wine.
You shouldn’t need to replace any hardware. If you have an Nvidia graphics card you will need to install the drivers as they don’t come with the kernel, but it will run just fine. I’ve heard of some issues regarding specific brands of headphones, and I had to fuss a bit to get my microphone and it’s audio interfacing working.
Adobe products, a lot of popular music production software and a few popular CAD programs will have issues. Most of them can be run on Linux, but they don’t like it, and finding an alternative would be better.
Imagine, that you were the intelligence service of a small country dependent on the support and compliance of larger countries to enable you to enact a genocide you had been planning for decades. What would be the best way to achieve that? Extensive Kompromat on influential individuals in positions of influence in media, politics and business might be one way to do that.
This isn’t a new thing ether, the lavender scare was in large part J “Edna” Hoover forcing queer individuals into a strict network of mutual blackmail. One of my grandparents told me several stories regarding the subject, don’t ask what they were doing in Budapest in 1956.
As has been made very clear, it is not actually possible to prevent these models from regurgitating any information they’ve been trained on, no matter how fancy the system prompt. So, if there is NSFW content In the training data, users will always be able to access it, not matter how “compliant” the company is with restrictions on NSWF content by way of system prompts.
They can have their cake and eat it to, many users will prefer the models because of their ability to do porn stuff, and they will not be held legally liable for that since they’ve done everything they possible could.
So long as no one proves that they did in fact intentionally train the models on a shit ton of porn …
I think some department or team lead was told by higher ups that they needed to “implement AI” and are just trying to find places to shove it so they can tell the boss they hit the metrics.
The reason they turn stuff like this on by default without asking creators or users is probably because previous things that were implemented and made optional all got turned off and not used, meaning it didn’t help the department or team hit their metrics for usage of the systems.
this guy was able to make a silicon chip with 1200 transistors on it in a garage, far from a modern CPU, but spitting distance from the 3500 transistors of the intel 8008 CPU. Projects in flight also has a series working on DIY semi conductor fabrication he’s not quite to the stage of fabricating a full chip, but he’s covered a lot of the more difficult parts already. And this is all stuff being done as hobby projects.
Something like an early pentium chip like the P5 would not be a particularly difficult task for a well resourced team. The questions are how much could they rely on the currently available off the shelf supplies and tools and how much do they need to scale production.
LDP, which has governed Japan almost uninterrupted for decades.
I think this actually undersells how unprecedented the LDP not being in power is. The LDP has been the ruling party in Japan since the end of US occupation in 1954, over half a century. The only interruptions being between 1994 and 1995 and between 2009 and 2011.
Japan’s modern poltical history is super interesting and less monolithic than that would make it sound, but it’s a big deal when the LDP isn’t in charge.
it’s so amazing, the absolute brain rot it takes to think that a LLM is a better way to operate a vending machine than simple if-then logic. “If the value of money inserted is equal to the price, then dispense the item”.
Like, why? What is even the point? It doesn’t need to negotiate the price, it doesn’t need have a conversation about your day, the vending machine just needs to dispense something when payed the right amount.