p3x.dediz, diz@awful.systems
Instance: awful.systems
Joined: 3 years ago
Posts: 8
Comments: 99
Posts and Comments by diz, diz@awful.systems
techtakes
Comments by diz, diz@awful.systems
Oh, by far. There’s only 80 decimal places in that at most.
It got to be a quantum sweatshop: a quantum computer for AGI (a guy instead)
How much does he think an engineer spends on CAD tools, anyway? Altium is like, what, $2500 / year? Very “how much can a banana cost”.
It’s all capital costs for tools, pretty much, anyway, maybe CAD should start charging per net lmao.
Oh they are going to charge per token for github copilot? That thing is a money waste for everyone, I’m pretty sure. I get a mix of inane mildly good suggestions, irrelevant stuff, and an occasional suggestion of super evil sabotage. Due to mild OCD about issues, I tend to have to fix said mildly good suggestions, but from the objective perspective that nitpickery is not worth it, everything was fine without, we had compiler warnings, coverity, etc.
edit: the difference being that the old stuff was deterministic and you just ran it on the whole codebase and had it pass. Unlike gh copilot that’ll just make up new shit. And as for the times it caught some bad bug that you made… add more tests instead.
I wouldn’t be too surprised if they really don’t, they’re just advertising the advertising lol.
edit: Basically what if you spent a trillion dollars so that you could beam ads to people’s bathroom mirrors. And better yet, ads reflected from water down in their toilets. Then in the interest of expediency you just take random ads and put them there for free, and your actual product, shares, sells better.
It makes every bad programmer into a 10x bad programmer (equivalent to 10 bad programmers).
I’m afraid they already had that exact idea when they named the startup “oklo”.
I think it’s not very difficult to construct a really shitty small reactor that is horrendously expensive per watt. Can probably be built in a year if you get rid of NRC and just half ass it completely.
I mean, Demon Core was a small reactor. You pretty much have to do a lot of work to ensure you won’t create a small reactor when a truckload of fresh fuel falls into a river.
What’s difficult is making a safe reactor that is actually making electricity at somewhat reasonable price per watt.
Nuclear already makes 9% of world’s electricity.
Curiously enough you also have those radon spas in old uranium mines. I don’t think hinging their luck on “a little radiation is good for you” worked for anyone.
Shorting the market requires precise timing. Being early is just as bad as being wrong.
Exactly. It is not enough to know that a company stock will go down. It is necessary to know that it will never go higher than a certain point above the current value (not even momentarily) before it goes down. If you have a fuckload of other people’s money you can just keep double-or-nothing-ing it, that’s what banks were doing to gamestop, except that this can sometimes cause the stock to go even higher (a short squeeze), which would make you (who doesn’t actually have a fuckload of other people’s money) lose all of your money.
edit: also the other concerning possibility is that stock prices can go up simply due to the dollar going down.
The only thing that is allowed to tell good art from slop is the AI which needs to consume good art and not slop.
Its spelled “masterdebating”.
Hyping up AI is bad, so it’s alright to call someone a promptfondler for fondling prompt.
I mostly see "clanker" in reference to products of particularly asinine promptfondling: spambot "agents" that post and even respond to comments, LLM-based scam calls, call center replacement, etc.
These bots don't derive their wrongness from the wrongness of promptfondling, these things are part of why promptfondling is wrong.
Doesn’t clanker come from some Star Wars thing where they use it like a racial slur against robots, who are basically sapient things with feelings within its fiction? Being based on “cracker” would be alright,
I assume the writers wanted to portray the robots as unfairly oppressed, while simultaneously not trivializing actual oppression of actual people (the way "wireback" would have, or I dunno "cogger" or something).
but the way I see it used is mostly white people LARPing a time and place when they could say the N-word with impunity.
Well yeah that would indeed be racist.
I’m seeing a lot of people basically going “I hate naggers, these naggers are ruining the neighborhood, go to the back of the bus nagger, let’s go lynch that nagger” and thinking that’s funny because haha it’s not the bad word technically.
That just seems like an instance of good ol anti person racism / people trying to offend other people while not particularly giving a shit about the bots one way or the other.
we should recognize the difference
The what now? You don't think there's a lot of homophobia that follows "castigating someone for what they do" format, or you think its a lot less bad according to some siskinded definition of what makes slurs bad that somehow manages to completely ignore anything that actually makes slurs bad?
I think that’s the difference between “promptfondler” and “clanker”. The latter is clearly inspired by bigoted slurs.
Such as... "cracker"? Given how the law protects but doesn't bind AI, that seems oddly spot on.
Note also that genuine labor saving stuff like say the Unity engine with Unity asset store, did result in an absolute flood of shovelware on Steam back in the mid 2010s (although that probably had as much having to do with Steam FOMO-ing about the possibility of not letting the next Minecraft onto Steam).
As a thought experiment imagine an unreliable labor saving tool that speeds up half* of the work 20x, and slows down the other half 3x. You would end up 1.525 times slower.
The fraction of work (not by lines but by hours) that AI helps with is probably less than 50% , and the speed up is probably worse than 20x.
Slowdown could be due to some combination of
- Trying to do it with AI until you sink too much time into that and then doing it yourself (>2x slowdown here).
- Being slower at working with the code you didn't write.
- It being much harder to debug code you didn't write.
- Plagiarism being inferior to using open source libraries.
footnote: "half" as measured by the pre-tool hours.
And yet you are the one person here who is equating Mexicans and Black people with machines. People with disabilities, too, huh. Lemme guess next time we're pointing and laughing at how some hyped-up "PhD level chatbot" can't count the Es in dingleberry, you'll be likening that to ableism.
When you're attempting to humanize machines by likening the insults against machines to insults against people, this does more to dehumanize people than to humanize machines.
edit: Also I never seen and couldn't find instances of "wireback" being used outside pro-bot sentiments and hand-wringing about how anti bot people are akhtually racist. Had you, or is it all second or third hand? It's entirely possible that it is something botlickers (can I say that or is that not OK?) came up with.
edit: especially considering that these "anti-robot slurs" seem to originate in scifi stories where the robots are being oppressed, whereby the author is purposefully choosing that slur to undermine the position of anti robot characters in the story. It may well be that for the same reason that author has in choosing these slurs, they are rarely used (in the earnest).
To be honest, hand wringing over “clanker” being a slur and all that strikes me as increasingly equivalent to hand wringing over calling nazis nazis. The only thing that rubs me the wrong way is that I’d prefer the new so called slur to be “chatgpt”, genericized and negative connotated.
If you are in the US, we’ve had our health experts replaced with AI, see the “MAHA report”. We’re one moron AI-pilled president away from a less fun version of Skynet, whereby a chatbot talks the president into launching nukes and kills itself along with a few billion people.
Complaints about dehumanizing these things is even more meritless than a CEO complaining that someone is dehumanizing Exxon (which is at least made of people).
These things are extension of those in power, not some marginalized underdogs like cute robots in scifi. As an extension of corporations, it already got more rights than any human - imagine what would happen to a human participant in a criminal conspiracy to commit murder and contrast that with what happens when a chatbot talks someone into a crime.
I think this is spot on. I had that same thing happen at my former employer, which bought a lot of entirely pointless startups in 2010s instead of investing in core business equipment and processes.
NotAwfulTechPython code really requires 100% branch coverage tests as an absolute minimim… with statically typed languages the compiler will catch some types of bugs in branches you don’t test, with python chances are it won’t.
edit: basically think of non covered lines the way you think about files you didn't compile.
Even to the extent that they are "prompting it wrong" it's still on the AI companies for calling this shit "AI". LLMs fundamentally do not even attempt to do cognitive work (the way a chess engine does by iterating over possible moves).
Also, LLM tools do not exist. All you can get is a sales demo for the company stock (the actual product being sold), built to impress how close to AGI the company is. You have to creatively misuse these things to get any value out of them.
The closest they get to tools is "AI coding", but even then, these things plagiarize code you don't even want plagiarized (because its MIT licensed and you'd rather keep up with upstream fixes).
One thing that I couldn't easily figure out is what is the constant factor. If the constant factor is significantly worse than for Strassen, then it would be much slower than Strassen except for very large matrices.
Let's say the constant factor is k.
N should be large enough that N^((log(49)-log(48))/log(4)) > k where k is the constant factor. Let's say the difference in exponents is x, then
N^x > k
log(N)*x > log(k)
N > exp(log(k)/x)
N > k^(1/x)
So lets say x is 0.01487367169 , then we're talking [constant factor]^67 for how big the matrix has to be?
So, 2^67 sized matrix (2^134 entries in it) if Google's is 2x greater constant than Strassen.
That don't even sound right, but I double checked, (k^67) ^ 0.01487367169 is approximately k.
edit: I'm not sure what the cross over points would be if you use Google's then Strassen's then O( n^3 )
Also, Strassen’s algorithm works on reals (and of course, on complex numbers), while the new "improvement" reduces by 1 the number of real multiplications required for a product of two 4x4 complex-valued matrices.
Old McDonald had a startup, iyo io o[4-mini].
It's funny how just today in a completely unrelated context a generative ai enthusiast used an example of OpenAI getting sued by NYT as a reason why they wouldn't commit some other malfeasance because they'd get caught if they did.
Its not about moats, it's about open source community (whose code had been trained on) coming out with pitchforks. It has nothing to do with moats.
Re-creating some open source project with a similar function is literally the only way a coding agent can pretend to be a programmer.
I tried latest models for code and they are in fact capable of shitting out a thousand lines of working code at a time, which obviously can only be obtained via plagiarism since they are also incapable of writing the most trivial code for a novel situation. And the neat thing about plagiarism is that once you start you can keep going since there's more of compatible code where it came from.
Yeah I'm thinking this one may be special cased, perhaps they wrote a generator of river crossing puzzles with corresponding conversion to "is_valid_state" or some such. I should see if I can get it to write something really ridiculous into "is_valid_state".
Other thing is that in real life its like "I need to move 12 golf carts, one has low battery, I probably can't tow more than 3 uphill, I can ask Bob to help but he will be grumpy...", just a tremendous amount of information (most of it irrelevant) with tremendoustremendous possible moves (most of them possible to eliminate by actual thinking).
Honestly 16 bit would be easier because there would just be so much less code.
Pre-LLM, I had to sit through one or two annual videos to the sense of “dont cut and paste from open source, better yet don’t even look at GPLd code you arent working on” and had to do a click test with questions like “is it ok if you rename all the variables yes no”. Ohh and I had to run a scanning tool as part of the release process.
I don’t think its the FSD they would worry about, but GPL especially v3. Nobody gives a shit if it steals some leetcode snippet, or cuts and pastes some calls to a stupid API.
But if you have a “coding agent” just replicating GPL code wholesale, thousands and thousands of lines, it would be very obvious. And not all companies ship shitcode. Apple is a premium product and ages old patched CVEs from open source cropping up in there wouldn’t be exactly premium.
That is not equivalent, though; other solutions to "can not be left unattended" exist; just ask Kristi Noem.
LLM snippets are so 2024. Coding agents, baby.
Yeah, that's a great example.
The other thing is that unlike art, source code is already made to be consumed by a machine. It is not any more transformative to convert source code to equivalent source code, than it is to re-encode a video.
The only thing they do that is "transformative" is using source code not for compiling it but for defrauding the investors.
Other funny thing: it only became a fully automatic plagiarism machine when it claimed that it wrote the code (referring to itself by name which is a dead giveaway that the system prompt makes it do that).
I wonder if code is where they will ultimately get nailed to the wall for willful copyright infringement. Code is too brittle for their standard approach, "we sort of blurred a lot of works together so its ours now, transformative use, fuck you, prove that you don't just blur other people's work together, huh?".
But also for a piece of code, you can very easily test if the code has the same "meaning" - you can implement a parser that converts code to an expression graph, and then compare that. Which makes it far easier to output code that is functionally identical to the code they are plagiarizing, but looks very different.
But also I estimate approximately 0% probability that the assholes working on that wouldn't have banter between themselves about copyright laundering.
edit: Another thing is that since it can have no own conception of what "correct" behavior is for a piece of code being plagiarized, it would also plagiarize all the security exploits.
This hasn't been a big problem for the industry, because only short snippets were being cut and pasted (how to make some stupid API call, etc), but with generative AI whole implementations are going to get plagiarized wholesale.
Unlike any other work, code comes with its own built in, essentially irremovable "watermark" in the form of security exploits. In several thousands lines of code, there would be enough "watermark" for identification.
AI solves every river crossing puzzle, we can go home now [content warning: botshit] (gemini.google.com)
I think this summarizes in one conversation what is so fucking irritating about this thing: I am supposed to believe that it wrote that code.
Having worked in computer graphics myself, it is spot on that this shit is uncontrollable.
I think the reason is fundamental - if you could control it more you would put it too far from any of the training samples.
That being said video enhancements along the lines of applying this as a filter to 3d rendered CGI or another video, that could (to some extent) work. I think the perception of realism will fade as it gets more familiar - it is pretty bad at lighting, but in a new way.
Well, it did reach for "I double checked it, I'm totally sure now" language.
From the perspective of trying to convince the top brass that they are making good progress towards creating an artificial psychopath - not just an artificial human - it's pretty good.
Still seems terminally AI pilled to me, an iteration or two later. "5 digit multiplication is borderline", how is that useful?
I think there's a combination of it being a pinnacle of billions and billions of dollars, and probably theirs firing people for slightest signs of AI skepticism. There's another data point, "reasoning math & code" is released as stable by Google without anyone checking if it can do any kind of math.
edit: imagine that a calculator manufacturer in 1970s is so excited about microprocessors they release an advanced scientific calculator that can't multiply two 6 digit numbers (while their earlier discrete component model could). Outside the crypto sphere, that sort of insanity is new.
Yeah, I'd also bet on the latter. They also added a fold-out button that shows you the code it wrote (folded by default), but you got to unfold it or notice that it is absent.
Oh and also for the benefit of our AI fanboys who can't understand why we would expect something as mundane from this upcoming super-intelligence, as doing math, here's why:
Also, I just noticed something really fucking funny:
(arrows are for the sake of people like llllll...)
lmao: they have fixed this issue, it seems to always run python now. Got to love how they just put this shit in production as "stable" Gemini 2.5 pro with that idiotic multiplication thing that everyone knows about, and expect what? to Eliza Effect people into marrying Gemini 2.5 pro?
The craziest thing about leaked prompts is that they reveal the developers of these tools to be complete AI pilled morons. How in the fuck would it know if it can or can't do it "in its brain" lol.
edit: and of course, simultaneously, their equally idiotic fanboys go "how stupid of you to expect it to use a calculating tool when it said it used a calculating tool" any time you have some concrete demonstration of it sucking ass, while simultaneously the same kind of people are lauding the genius of system prompts half of which are asking it to meta-reason.
Thing is, it has tool integration. Half of the time it uses python to calculate it. If it uses a tool, that means it writes a string that isn't shown to the user, which runs the tool, and tool results are appended to the stream.
What is curious is that instead of request for precision causing it to use the tool (or just any request to do math), and then presence of the tool tokens causing it to claim that a tool was used, the requests for precision cause it to claim that a tool was used, directly.
Also, all of it is highly unnatural texts, so it is either coming from fine tuning or from training data contamination.