The question is, however, is it really "open" if it refuses to address well known and documented facts? And facts that it was clearly exposed to. Doesn't scream "openness" if you ask me. It's not like it's social credit could go down, if it was truly "open".
I love how people love throwing the word "left" as it means anything. Need I remind you how many times bots were caught on twitter using chatgpt praising putin? Sure, go ahead and call it left if it makes you feel better but I still take the European and American left over the left that is embedded into russia and china - been there, done that, nothing good ever comes out of it and deepseek is here to back me up with it's answers.
So long as you don't ask it about tiananmen square 1989. Or Tibet. Or Taiwan. Or the Xinjiang internment camps. Just a few off the top of my head but thousands of others if you decide to dive deep. You get a shrug at best. Which does beg the question what responses you'd get in certain contexts.
I have to disagree here. AI tools are great and work great. Until they do not. The speed with which they spit out a result gives a great sense of confidence and you may overlook an edge case which they do not cover. And it's not a question of if but when you will find out that you've crashed a million systems worldwide or made the wrong update on a 10TB table which ran for 3 days. I almost did the latter just yesterday because I was lazy to write a seemingly dumb and trivial function. Luckily I am pedantic enough to quadruple-check before I hit enter.
Well yeah... But overall numbers in French are a bit... Weird... By the time you are a teenager, you have the math skills of someone with a PhD in Calculus just to be able to say how old you are. No wonder some of the best mathematicians in history were French ;)
I speak Spanish, but it's a different story there: "dos doses, zero treses y un cinco". Numbers can have plurals, which from what I understand is not the case in Italian. Weird cause the languages are very similar in general - I can somewhat easily understand Italian, particularly reading. Listening - not so much. But as far as grammar, they seem to be almost identical. Same with French grammar though Spanish has the equivalent of the English present continuous tense and French does not(also worth mentioning that I don't speak French either, that's what my mum has told me).
I am not sure that it's a function of English language per se. I speak several language and it's the same story with all of them and one of those languages is Slavic so it comes from a very different root. That said, Greece is a rock throw away and I think the ancient Greek mathematicians(Pythagoras primarily) might have something to do with it: The Egyptians were the first known to use symbols to represents parts of something but it wasn't until the Greeks introduced fractions to express a quantifiable representation of sub-divisions of a unit, making the sub-division it's own unit: you need 4 * 1/4-th's of something to make it to 1 complete unit.
Bambu lab printers are truly awesome in terms of what they can do for a very reasonable price. Having said that, I have never upgraded mine nor have I ever connected it to the internet and never will. Nor will I update it. If it takes me 15 minutes to get an ssh client running on an esp8266 that can connect to an
poorly secured server and execute shell commands, there is no way I'm letting a proprietary chinese hardware and software anywhere near my home network. But this is just a side hobby of mine, so I can live with carrying around an SD card. But I can see how something like that can be a major blow to business owners. I am not entirely sure if this blog post is just a response or sneaky backpedaling from bambu labs after the backlash they received over the last few days.
* The software: different open source solutions have very different requirements at a high level: language, platform or even system requirements. Say you want to take messaging off centralized platforms: you need to host something like Matrix, which is very well made and polished but takes a lot of resources to run. Alternatively, you could use Jabber, which scales like no other but is an absolute hell to setup and maintain. Same can be said about music, videos, movies and all other things
* Operations: probably simple if you ask someone on HN, but you still need to understand networking, operating systems and file systems. I started using Linux when I was 11 in the distant 2000, and even now I'm not very enthusiastic if I have to make some changes to my zfs. You also need to consider backups and security and resources. Say you wanna run openstreetmap(which we recently started doing at work). Awesome but that requires an ungodly amount of fiddling in addition to an astonishing amount of time needed to unpack, even on enterprise hardware.
If you are in the tech world, https://github.com/awesome-selfhosted/awesome-selfhosted is a great place to start. But if you want to make it simpler... Idk... A lot of people would need to put in a lot of effort, as in build a linux distro around this idea, along with "recommended hardware", one click install(a very dumbed down equivalent of portainer), and some backup and alerting mechanisms built into the system. It's a tough question and frankly I don't have the answer.
Our age gap is less then 10 years but here's my two cents: laziness/convenience. Back in the 90's and 2000's, you had to be ready to spend a lot of time fiddling with setups and maintenance as well as some MAJOR early days security flaws(think the IRC days). Corporate-owned ecosystems solved that problem: you log in and forget about it. They won with what some people call user experience. The lower the entry barrier, the quicker something picks up. Back when I was in school I was the biggest Nokia fanboi and even then I acknowledged that downloading a shady jar file and installing it on my phone was iffy. At a later stage when I was a bit older and could afford it, I got my first Android phone and the existence of a marketplace was a breath of fresh air. The problem is that few people(annoyingly even now) fail to realize or admit that those types of centralizations put handcuffs on your wrists the moment you say "OK, that works for me". Whether that's social logins, cloud providers, services or anything else - it's all the same. For example, if today, OpenAI decided to close off their API's for good, I recon tens if not hundreds of thousands of "AI" startups will collapse immediately since they fully rely on OpenAI's API's. Same with AWS, GCP, Azure or any other provider. And as we see with the current fiasco with twitter, tiktok and bambu labs just to name a few from the past two days, it is abundantly clear that people are in dire need of backups. As much as I used to find google drive and docs convenient, I've personally moved away and self-host everything now. The only thing I rely on(and only as a backup plan to access my home network) is a VPN I host over at Hetzner. But again - this is my backup.
Whether the corporations saw that as an opportunity at an early stage or they were at the right place, at the right time, I can't say. I'm more leaning towards the latter since I've worked at corporations and success in those environments is most commonly a moderately-educated gamble.
Seems I spoke too soon about the US taking a good decision for once when it comes to cyber and civil security. Well... I wonder what muskov will come up with now that twitter is still at large inaccessible in China but tiktok is welcome in the US.
Historically the US has been always behind the curve when it comes to personal online privacy and security threats compared to Europe(and it's not like we are doing a good job at all, Europe is pretty shit as well), but here, today, I am utterly jealous of the US. Let's hope the EU commission gets their shit together and follows soon.
Yes, how did I not think of throwing more money at cloud providers on top of feeding open ai, when I could have just code a simple binary classifier and run everything on something as insignificant as an 8-th geh, quad core i5....
Ok, let me dumb it down for you: you have a cockroach in your bathroom and you want to kill it. You have an RPG and you have a slipper. Are you gonna use the RPG or are you going to use the slipper? Even if your bathroom is fine after getting shot with an RPG somehow, isn't this an overkill? If you can code and binary classifier train a classifier in 2 hours that uses nearly 0 resources and gives you good enough results(in my case way above what my targets were) without having to use a ton of resources, libraries, rags, hardware and hell, even electricity? I mean how hard is this to comprehend really?
FFS... "Lots of writers, few readers". Read again and do the math: 2 seconds, multiply that by 10 million records which contain this, as well as "alarm installation in two locations" and a whole bunch of other crap with little to no repetition (<2%) and where does that get you? 2 * 10,000,000 = 20,000,000 SECONDS!!!! A day has 86,400 seconds (24 * 3600 = 86,400). The data pipeline needs to finish in <24 hours. Everyone needs to get this into their heads somehow: LLM's are not a silver bullet. They will not cure cancer anytime soon, nor will they be effective or cheap enough to run at massive scale. And I don't mean cheap as in "oh, just get openai subscription hurr durr". Throwing money mindlessly into something is never an effective way to solve a problem.
No, it has not and will not in the foreseeable future. This is one of my responsibilities at work. LLMs are not feasible when you have a dataset of 10 million items that you need to classify relatively fast and at a reasonable cost. LLMs are great at mid-level complexity tasks given a reasonable volume of data - they can take away the tedious job of figuring out what you are looking at or even come up with some basic mapping. But anything at large volumes.. Na. Real life example: "is '20 bottles of ferric chloride' a service or a product?"
One prompt? Fair. 10? Still ok. 100? You're pushing it. 10M - get help.
That IS the second system (my AI home rig). I've given up on Nvidia for using it on my main computer because of their horrid drivers. I switched to Intel ARC about a month ago and I love it. The only downside is that I have a xeon on my main computer and Intel never really bothered to make ARC compatible with xeons so I had to hack my bios around, hoping I don't mess everything up. Luckily for me, it all went well so now I'm probably one of a dozen or so people worldwide to be running xeons + arc on linux. That said, the fact that I don't have to deal with nvidia's wretched linux drivers does bring a smile to my face.
> they seem to be doing everything right in the last few years
About that... Not like there isn't a lot to be desired from the linux drivers: I'm running a K80 and M40 in a workstation at home and the thought of having to ever touch the drivers, now that the system is operational, terrifies me. It is by far the biggest "don't fix it if it ain't broke" thing in my life.
To be honest it is still a problem till this day in some areas. A few years back I was making a gift which involved a ton of electronics and embedded hardware. One of the components I had laying around was a real time clock which refused to go beyond 31.12.1999, no matter what I did. Turned out that in (2021 iirc), there were still rtc modules on the market that had this problem. But it did not affect me all that much since the date had to be displayed on screen so I did the hacky-patchy thing to trim the first two bytes from the string and prepend "20". I bet there are tons of software and hardware out there that use dirty hacks like that to "fix it".
I'm really glad to see Geerling branching out of his initial niche coverage on everything-raspberry-pi(which I personally find a bit boring 95% of the time).
Most likely an effort to boost the DAU numbers. I quit facebook over a decade ago because I truly felt that it was pointless. At the time I was convincing myself that it's a way to stay in touch with a certain number of people I would otherwise have no way of contacting. Then a friend said something that changed my mind: "If someone is not actively a part of your life, chances are there's a good reason they are not". And he was right: I deleted it, knowing full well that I'd have no other way to connect to hundreds of people. Over a decade later I haven't had the reason to try and contact either one of them. At this point, I don't even know what facebook looks like but all this AI-generated crap is just as pointless as the ads that would get shoved down my throat if I didn't have a very aggressive ad blocker. As much as I was strongly against ad blockers 10 years ago, since many sites and blogs used only that to get some reward for their effort, we are at a point where the internet is unusable without an ad blocker. All major platforms are flooded by AI-generated crap. And I mean Facebook, Medium, StackExchange, hell, I'm willing to bet a good chunk of papers coming out these days are mostly ai-generated. And don't even get me started on musk's shithole that is Twitter. No, I am not saying that AI is not useful or helpful - it is, but it should be a supplement, not the primary ingredient, let alone the sole ingredient.
The true value of the internet used to be the collective knowledge, and not mass-produced regurgitated set of tokens and pixel values. Personally I've gone to the even pre-rss days and have a list of personal blogs I scroll through for things I find interesting and avoid large platforms altogether. Interestingly enough, I've been finding more and more motivation to start writing myself though I rarely get the chance to push it through the end and in most cases I get stuck at 95% for many months until I get to find the time to do the remaining 5% of the work. That's how many I have lined up so far:
Given that Denmark gave Ukraine F-16s and didn't bat an eyelash, I'd be all for it. I would be pretty happy if they hand out 2-3 fully loaded B-2's and be like "hey guys, go nuts with them".
Delivery success rate is pretty much a random number generator here. Sometimes stuff arrives on time, sometimes it shows up 6 months late, sometimes never. For me it's either the physical store, amazon or mouser.
I think the one thing no one really mentions is the fact that drones are extremely easy to build in the comfort of your living room. If you have a 3d printer, some spare parts laying around and a cheap electronics store around the corner(check, check and check in my case), you can build a drone for less than 300 bucks which is far less than the retail ones - completely autonomous, no radio link needed, takes off, does what it does and lands on it's own with no human intervention.
This is a question of definition. It is synthetic in that I just passed a prompt, asking for N amount of examples of X. And I did not go over the entire list I got and blindly trusted it. In this context, I needed an even(or nearly even distribution) of samples in the training data and it worked way better than I was hoping. Mind you, I have to face a similar issue next week and I'm not sure this approach would cut it - I need way more training data and way more classes to work with. 249 classes if I'm not mistaking.