Posts by tekHedd 700 publicly visible posts • joined 9 Aug 2011
He's right and he's wrong.
Wrong: MySQL is totally irrelevant--if, previously, you needed MySQL and not Oracle, you now need MariaDB and not PostgreSQL. (Or possibly just a sqlite embed.)
Right: If you needed the scaling and features of Oracle before, you are SO on Postgres now, even if the boss fought you for YEARS on the move. (Ask me how I know.)
The only thing that changed is that now "nobody gets fired for not choosing Oracle".
Seeing the headline, my first thought was that it is just an article from 15-20 years ago that got accidentally reactivated.
Screen readers? Heck a generic mouse-keyboard full-vision person has to experiment with each new site to figure out how to navigate the most basic things. It's almost as if two generations of web designers never bothered to read even a short whitepaper on UI design, or even look at another site to consider what would be the "least astonishing" thing.
Stop Innovating.
"AI"? Not even. "AI" isn't here and never happened. But, it's 5 years out--like cold fusion.[*]
Until AI happens, we have the pipe dream: LLMs. These will continue to be a sometimes-useful irritant because management will *never* give up on trying to replace people with anything that remotely resembles a replacement, and the people you get when you aren't willing to pay them enough to live are so bad at their jobs that this almost works. This isn't really transformative, it's more akin to the huge game changer of "touchscreens in cars" or "every phone has a 50 megapixel camera that you use to take pictures of your lunch."
Neural nets are really cool and can model all kinds of fun stuff! They're pretty good at categorization. They're going to stay with us now that we have the tools to program them. This change will not be visible.
The point is: the technology was never the problem.
(* - we call it "AGI" because General Artificial Intelligence, the order you would normally use, spells "GAI". Perhaps it would be less reluctant to materialize if we were less afraid to say its true name.)
Hmm, so multiply by 10x or 20x to get the "actually profitable" pricing and considering (based on the article) that it's doubling the LOC while producing an unmaintainable mess... it's really not that impressive, is it? How long would this job have taken on my local computer with a 4090?
Yeah ID checks "could" turn into an ID checkpoint the same way Flock and Ring cameras "could" be used as a Big Brother style surveillance network with no oversight, government spy agencies "could" tap into the central internet backbones and phone networks and archive all interesting traffic for later searching...
But surely that won't happen, right?
The claim is that the US engaged in cyber attacks, and *only* targeted backdoors in US made equipment? Now THAT is what I call BS. Obviously we would have used any exploits we know about, including those in Chinese-made equipment. Is there perhaps some doubt that the NSA knows where the backdoors are? That they don't have a stash of 0-days in reserve?
We might have hypocritical propaganda, but we're not stupid.
Lots of downvotes here, but it's worth considering that one of the primary powers of the executive branch of government is selective enforcement. Having a lot of laws that are widely ignored but can be enforced on discretion gives the police a whole lot of power. If you're comfortable with that, well, I suggest you come visit America this spring, and be ready to unlock your phone.
You can try to "spark a debate," or you can just assign them different names and stop using the old ones, maybe publish a lookup table for reference...
https://gayint.org/actors/public
Maybe install the browser plugin that replaces all the names with their proper GAYINT actor names so you don't have to look them up all the time...
Or yeah we can discuss the merits of doing so for another year that's totally an option. ;)
Systems like my gaming PC, that sat idle for a month because I really didn't have time to game, got no updated anything other than automated Windows updates, and stopped booting Windows 11 after the latest update. Could not rollback or recover. The drives are fine.
It's now running Arch.
Well, with the ubiquity of AI Slurp[TM], I think the "Open" part of open source is pretty much doomed. If you don't want your code stolen outright and then sold to all comers on a metered basis, you don't open your source.
So now we have two good reasons to close open source... but that's not really a *good* conclusion, it's just possibly unavoidable.
File under "why we can't have nice things" I guess.
This is an attempt to slander AI criticism by equating it to the widely-mocked Veganism. (Hey I was vegan for 3 months, never felt so healthy or so HUNGRY.)
Title of the piece implies that it is friendly to AI skeptics. It is not.
"Skeptics." Oh look, there *is* already a word for how I feel about LLMs being a silver bullet. I didn't have to look at all.
Electric space heaters are an interesting exception to the rule, as even the cheapest ones are 100% efficient. What happens to the waste energy? It turns into... heat.
Turns out, they're expensive to run not because of efficiency, but because it's actually kind of difficult to create and move electricity. It's very easy to turn it back into heat.
I've been describing LLMs as "just the entire internet stored with lossy compression." All LLM "deficiencies" are side effects of this.
This article describes what is lost if you tune the model to mask IP theft, or don't have a large enough model.
At a high enough quality factor, almost no data is "lost" in the lossy compression, and the similarity between "AI" and outright theft becomes visible. Introducing randomness to Gen AI is not a feature, but rather a hard requirement if they want to continue pretending it's not just massive theft of IP while making the models larger.
Hallucination, of course, is the only thing AI does, but also an inevitable side effect of the compression algorithm. AI can interpolate between the stored (stolen) data to generate intermediate or extrapolated responses, which is useful (in an artistic context) or harmful (if you're looking for medical information).
It's just crappy data compression.
"...questioned whether this technology could be used for more nefarious means."
Questioned?
QUESTIONED?
Don't be silly. Nobody questioned it. The people pointing out how obvious the nefarious uses are were just being polite. Some people just don't know what a dystopia is and/or honestly don't believe that technology can be used for evil.
Gen AI is lossy compression. As the model becomes large enough, it approaches "not being lossy" and the primary business model is "hiding the fact that it's just the entire internet loaded into memory as an algorithmic model."
Which is to say, it's held back by the need to pretend it's not theft.
Conclusion: Gen AI can achieve its goals by removing ethical constraints. Isn't this how the System Shock game starts? ;D
Exactly this: "trusted" computing is one of those inverse names. What it really means is "you are no longer trusted."
Adding this to Linux is just another way to take away control over your OS and your personal computing. SystemD was a fairly obvious and successful play to take Linux away from small developers. This is a move to take control from the end users.
Obvious power play is obvious.
Yes, you WOULD expect better from a big company like Logitech... unless you'd used any of their included software apps in the last 25 years. Which is sort of sad. They make arguably some of the best hardware in the world, and it's supported by some of the worst apps.
I remember their keyboard mapping app written in .NET, that used a bunch of ram, so it would swap out of memory while you were playing a game and cause lag. The mouse apps that would... oh never mind you get the idea. It's just that Logitech being pants at app development is nothing new.
Exactly! We're approaching something. If you redefine "the singularity" from its previous apocalypse-adjacent meaning[*] to mean "the absolute best result as an LLM can ever hope to achieve," then we're basically approaching it, and always will be, because it's an asymptote. We can put exponentially more processing power into an LLM, but the training data is finite.
"AI" of today asymptotically approaches being a perfect search engine that makes stuff up if the answer was never posted to reddit. And it has no idea whether the answer is real or made up, so it can't even warn you.
The "apocalypse" analogy runs deep. The singularity will always be just around the corner.
* singularity: the point in time when machine intelligence reaches to the point where human intelligence is no longer relevant and change happens so fast that the speed of development is effectively infinite, such that the rate of technological progress is nonlinear in the seminconductor sense, similar to reaching the event horizon in a black hole
I tried a demo of ArcGIS on a project (converting ancient map data for my HOA) and it was, to put it nicely, complete garbage. A bag of hackish python scripts with no error detection or reporting. They still send me a beautiful, thick magazine full of pictures every few months. I guess we know where all that sweet government money goes: glossy brochures.
I'm not surprised it has a few holes in it.
Oh no, it's TEN YEARS old! Might as well set fire to it. Bose lost the plot when old man Bose passed on (as he apparently foresaw). Even this article suggests that it's reasonable to expect to stop using *speakers* after 10 years.
Ridiculous. Streaming radio hasn't gone away. Spotify, last I checked, is still in business. The hardware doesn't wear out. The speakers in my front room are a 30 year old pair of Definitives (Craig's list), powered by a 50 year old Carver and a $11 bluetooth receiver. Last I checked, WiFi was still backward compatible with 2013 standards. Firmware is updatable.
Everything these days is built to a 4 year refresh cycle, like American cars from the '90s. It seems like a good idea from a corporate standpoint, but it really kills the brand. Example: UA Project Rock headphones by JBL: replaceable earpads! But you can't replace them because they stopped making the pads at 4 years when they stopped making the phones. (And it's actually worse than other brands because they use a proprietary attachment mechanism instead of a simple stretch-on pad.) "Repairable" is meaningless branding now, like "green" or "washable."
We complained about planned obsolescence in the 90s, but those companies lacked vision.
It's the "some speech *should* be illegal" argument. We've used it here in the good old USA for years to explain why some objectively bad speech should be illegal. Now an objectively horrid party is in power and now they're using that on what *they* consider speech that should be illegal.
So, I mean, I'm not going to disagree with you (who wants to become unemployable); I am just pointing out that this particular sword has two edges.
The title of this article demonstrates that not only do the authors not know how neural nets work and are trained, but neither do the OpenAI researchers looking into "halluncinations."
Of *course* it rewards making stuff up, making stuff up is the *only* thing generative AI does.
Somebody gets paid to do this research? Paid, probably, like 5x what I make? I would take that job but I'm not good at making stuff up.
I'm still amazed that nuclear is considered dirty power while gas turbines fed by "let's loosen those regulations a bit more" fracking is considered green.
Yes we all learned that nuclear waste was horrid in the 60s, but energy companies have ridden that wave for a lifetime. Things have changed.
Wait, so just one of these GPUs pulls as much power as my house? (Well, not *my* house. I have a lot of crap in here, but a normal person's house?)
Sounds about right.
After the burst, we'll be able to rent this excess capacity, but it will still use a shedload of power, which is not getting cheaper. And the next generation of chips will be more efficient. So, just as developers in the suburbs are building the slums of the future, these AI companies are filling warehouses with e-waste and way more power and cooling capacity than is actually needed.
Smart money is considering how to make money with all that power and cooling capacity, once the obsolete GPUs have been carted away.
Maybe they think that invoking the (undoubtedly real) specter of "China" will protect them, but if they're a US government agency, perhaps they should keep their heads down right now. We're still a very long way from funding those tax cuts for the rich, and any functioning government agency is a target.
I know, that's really cynical. These are cynical times here in the good ol'.
I mean, if you accept that widespread adoption of IPv6 and the end of IPv4 is a good thing, this plan is good and I see no problem with it.
We may safely ignore IPv6's detractors, because its simple elegance makes it a no-brainer for... it's simple easy-to-understand ... its. I'm sorry where was I going with this?
Anyway, if you want to drive adoption, just make the use of IPv4 anywhere in the system a hard incompatibility with systemd. This will of course first require moving the rest of the network stack into systemd, which I assume is already on the roadmap. This will both win hearts and minds from the hardcore Linux developer community, while gently moving the backbone of all computing architecture to IPv6.
