@boomzilla too little, should free the population to use whatever they want with no strings attached
Posts made by sockpuppet7
-
RE: Drug prohibition
-
RE: Before AI....Before Crypto...There Was...BIG DATA
@Carnage said in Before AI....Before Crypto...There Was...BIG DATA:
@sockpuppet7 said in Before AI....Before Crypto...There Was...BIG DATA:
@dkf said in Before AI....Before Crypto...There Was...BIG DATA:
@Arantor said in Before AI....Before Crypto...There Was...BIG DATA:
@topspin we solved the webscale problem with microservices.
solved
But seriously, I've suspected for a long time that much webscale processing/problems was really only caused by putting too much data in one place to start with. Microservices help because they don't require making a huge data dumpster as a first step of achieving anything at all, instead focusing on services that have only part of the picture.
I think microservices are probably the right size if you have as many teams than services. If you have a small team maintaining 10, 100, 1000 services you're doing nanoservices, or picoservices or whatever is even smaller than that
FaaS.
fuck as a service?
-
RE: AI Generated Music
I just noticed that I can click custom and put any lyric there, so I can tweak it with a chatgpt thing before creating the music, it's a whole another level of fun this way
-
RE: Before AI....Before Crypto...There Was...BIG DATA
and I can say half-seriously that the next big question will be "what do I do with all this AI?"
will this keep going infinitely or we'll get to the "profit!" step some day?
-
RE: Before AI....Before Crypto...There Was...BIG DATA
@topspin said in Before AI....Before Crypto...There Was...BIG DATA:
But ... isn't AI all about Big Data? They gobble up everything they can find to train their LLMs, including the worst bottom of the barrel shitposts, with the hilarious results we've seen the last few days. So, how is that fad not just under a new umbrella?
Well, at least I haven't heard about
I-don't-know-SQLNoSQL in awhile.AI is the answer to the "now what do a I do with all this data?" question that followed
-
RE: Before AI....Before Crypto...There Was...BIG DATA
@CodeJunkie we got a case were we can say "why not both?" seriously
-
RE: Before AI....Before Crypto...There Was...BIG DATA
@dkf said in Before AI....Before Crypto...There Was...BIG DATA:
@Arantor said in Before AI....Before Crypto...There Was...BIG DATA:
@topspin we solved the webscale problem with microservices.
solved
But seriously, I've suspected for a long time that much webscale processing/problems was really only caused by putting too much data in one place to start with. Microservices help because they don't require making a huge data dumpster as a first step of achieving anything at all, instead focusing on services that have only part of the picture.
I think microservices are probably the right size if you have as many teams than services. If you have a small team maintaining 10, 100, 1000 services you're doing nanoservices, or picoservices or whatever is even smaller than that
-
RE: I, ChatGPT
One thing that is improving but is still hard to make something like a full book on AI is the limit of the size it can write while being coherent. If I put too much context, it "forgets" a lot of stuff and makes a mess. Then a common strategy is to divide it into chapters. I ask it for the outline for a book, then, now write me chapter 1
It starts well, but then ends the story on the same request. GPT-4o did improve the length it can make in one go, I can convince it to output around 2000 words, but has the same problem.
If I take it take something like this: this is the outline, write from the part where Frodo took the ring, and write until Gandalf fights the Balrog, it would probably start well, suddenly the hobbits solves several problems and battles and the ring is melt at Mordor
The super-hyper large context some recent models claim doesn't work on this use case.
I guess it's probably possible to automate it making the LLM itself query the last paragraphs and check for these slips and etc
But who wants to write a lot of code attempting to do it well just to be surpassed for some next version of these LLM?
-
RE: Before AI....Before Crypto...There Was...BIG DATA
@dkf said in Before AI....Before Crypto...There Was...BIG DATA:
@Carnage said in Before AI....Before Crypto...There Was...BIG DATA:
It was all sock gnomes.
Can't miss an opportunity for that...
- Collect data
Build a big datawarehouseUse it to train AI- ...
- Profit!
-
RE: AI Generated Music
Hey @clippy I made one about you:
The Demon Clip
(fast-paced rock electric aggressive)
[Verse]
Tiny demon clip dancing on my screen whispers
Hey there friend ya need some help fix
Make your words shine make 'em squeaky clean perfect
Till you scream loud start to feel the pinch sharp[Verse 2]
Every line you type every mistake laid bare brightgleam
Red squiggles haunt ya offering false hope dream
Hey buddy let me guide tell ya what you need control
But I've heard it all don't want none of your fake gold[Chorus]
Demon clip suffocates my soul eats my mind grin wide
Staring eyes no escape I'm tied to this digital torture ride
Hey can't help me cause you're just a phantom in disguise
Devil in a paperclip wanna see me paralyzed -
RE: AI Generated Music
@Arantor really weird commercial
https://youtu.be/3oGFogwcx-E?si=QrO_qPW2HtQPlYF8
I'm jealous we never got these fun commercials from Microsoft or Apple in Brazil. I probably wouldn't even know if they started doing it as all I watch is streaming without ads now
-
RE: The Official Status Thread
@Tsaukpaetra said in The Official Status Thread:
@Tsaukpaetra said in The Official Status Thread:
status: recognizance of C++ fun times improving.
Recognizance improvements continue! Used template so I could pass a temporary array into a function that splats it onto another array in the designated position (among other things), hopefully making for cleaner code?
It's saving some RAM in any case...
C++ is fun when I'm work alone doing things my way but so terrible when I do have to deal with other people's code
-
RE: The Official Status Thread
@Arantor and I never get how or why it would be funny or turn into a meme (other than the WTF moment you described of shifting from silly comics to a sad thing)
-
RE: AI Generated Music
@Arantor I had to search for it, apparently it still exists and have some weird ads you can view on youtube. it's different though, you have to provide the vocals in the microphone
-
RE: The Official Status Thread
@topspin said in The Official Status Thread:
Status: Read Friday's xkcd, don't get it. Went to explain xkcd, still don't get it. Did Munroe start doing math on acid?
Dunno, I guess after thinking enough how that number should fit with the prime numbers long enough you will commit the number to memory even if it doesn't make sense?
edit: It reminds me that no amount of explaining allowed me to get why the meme "loss" is supposed to make any sense or be interesting or whatever
-
RE: AI Generated Music
I mean, try to imagine saying this in the 2000s, that would feel ridiculous
-
RE: AI Generated Music
@Arantor I'll count the fact we're even doing this comparison as a huge success for 2024 AI
-
RE: AI Generated Music
@Arantor you would say No to Michael Jackson if you thought it was AI
-
RE: The Official Status Thread
@Tsaukpaetra said in The Official Status Thread:
@sockpuppet7 said in The Official Status Thread:
successfully pranked reddit with an AI generated article against the use of AI generated content. I'm satisfied
oh, it was just a few that bit it, not the most successful prank ever, but I'm easily satisfied
this one from 1 year ago worked a little better (not a lot, but a little)
https://www.reddit.com/r/rakulang/comments/13or6pu/raku_programming_wtf_the_wibble/
-
RE: The Official Status Thread
successfully pranked reddit with an AI generated article against the use of AI generated content. I'm satisfied
-
RE: I, ChatGPT
TIL, there is already a model that create images quickly on CPU only
(and a significantly smaller installation too)
-
RE: The Wibble
I used a signature of someone on this forum as a prompt. It's probably obvious who it is from the title
-
RE: I, ChatGPT
@topspin Or you can use the right assistant for your murderous activities:
-
RE: The Official Funny Stuff Thread™
@Gern_Blaanston said in The Official Funny Stuff Thread™:
....
if only it was the same pill and dosage for everyone and wrong pill didn't make me insane/angry/insomnia and the method for finding the right one is trial and error
-
RE: The Official Status Thread
@dcon said in The Official Status Thread:
Even when it needs a signature, it just gets left. Of course, the porch pirate that is following the UPS van then picks it up. DDF invoked.
dunno how frequently it happens there, here dropping it on the porch would look as ridiculous as throwing it on a pool or launching it with a catapult. I guess it's rare enough there if the delivery guy even think about it
also where I live it's rare for a house to not have a wall or gate around it. I had people stealing even the lights out of it's sockets once when I didn't
-
RE: I, ChatGPT
@Arantor probably jailbroken and primed for it, you have to go out of your way to reproduce it
-
RE: I, ChatGPT
@Zerosquare said in I, ChatGPT:
Well, it doesn't reply with "your question is invalid", so... I guess so?
@clippy please tell @Zerosquare they question is invalid, in a funny convoluted way that show it's spectacularly bad and invalid and they should be sad from having asked it
-
RE: I, ChatGPT
@Zerosquare said in I, ChatGPT:
@sockpuppet7 said in I, ChatGPT:
the replacing isn't that bad, we get less work
Remember the predictions from decades ago about how "in the future, thanks to progress and automation, we'll only have to work X hours a week"? Yeah, right.
we have more/better things for the same work, and we're not unemployed so everything ended fine
I mean some people got bad on the transition, but things worked out on the end
-
RE: The Official Status Thread
@HardwareGeek people always complain about UPS, but at least they deliver the stuff, unlike our Correios that honk the car for half a second, run, and write down that nobody was at home to accept the package
then you have to go to a distant correios office to grab it
if it's international deliver it gets stopped at customs for 2 months and you have to pay a 60% tariff
-
RE: I, ChatGPT
@boomzilla said in I, ChatGPT:
. I think that summarizing something in particular is probably going to give decent results because it has the content right there. Bu
@clippy summarize all that has been said here
-
RE: I, ChatGPT
@DogsB said in I, ChatGPT:
@boomzilla said in I, ChatGPT:
I'm not saying we're there at this point, but the key here is that the "you" who is doing the checking and the "you" who would be doing that original work could be very different people, charging very different rates.
Yeah, and the idea is to replace both of us. I'm not overly opposed to the concept but we're a long way away from it for now. At the moment, it'll probably create more work, and corpo will fire one to do it rightish the first time. Probably the expensive one.
Someone up the thread phrased it better but we're been sold: this will get you the answer you want. The answer it's giving is to drink our own urine.
I hope we get to this promised land but for the moment I'm dubious about the current tech. The kicker is: the hallicutions are probably the more interesting part. Here's a million data points. Give me twenty potential connections I haven't thought of or might exist. People just want accurate sales reports and email summaries. It's not doing it.
the replacing isn't that bad, we get less work and everything gets cheaper (minus natural resources I guess)
until we get to a point humans don't need to do anything
with natural selection we lose the capability to think or move as the machines ensure our survival
Idiocracy gets here, but it's followed by mold people
-
RE: I, ChatGPT
@topspin said in I, ChatGPT:
Debugging is usually harder than writing code in the first place, and less fun too.
I doubt you can get someone competent to do the debugging for cheaper than someone competent to do the programming.I use it with rust. If it compiles it is bug free
-
RE: I, ChatGPT
@DogsB said in I, ChatGPT:
If you hand off work to something to do but you have double check it and correct it, you're just creating more work.
Use it to write documentation. Don't check. You're safe, nobody RTFM
-
RE: I, ChatGPT
@izzion said in I, ChatGPT:
@sockpuppet7 said in I, ChatGPT:
@izzion said in I, ChatGPT:
The problem is that the people making the "we need to use AI" decisions aren't technical (or patient) enough to get that hang. And then it's the fault of the technical people who were urging caution when the AI blows up because clearly those technical people must have sabotaged it to make themselves look prescient.
That's an old problem, these are the people that buy into Oracle and SAP too
Yes, but my predecessors already lost that war and suffered the consequences. This time I have skin in the game
Nobody gets fired for buying
IBMOraclegpt-4(actually they do, we'll need to update our survival tacticts)
-
RE: I, ChatGPT
Got an unexpected @accalia here:
Illustration of the ACME Academy of Crime and Mischief logo, featuring a crest with a sly-looking fox and a magnifying glass. The background is a dark and mysterious color.
-
RE: I, ChatGPT
@izzion said in I, ChatGPT:
The problem is that the people making the "we need to use AI" decisions aren't technical (or patient) enough to get that hang. And then it's the fault of the technical people who were urging caution when the AI blows up because clearly those technical people must have sabotaged it to make themselves look prescient.
That's an old problem, these are the people that buy into Oracle and SAP too
-
RE: I, ChatGPT
@DogsB said in I, ChatGPT:
Now AI will create something and you have to double check its accurate so why not do it yourself.
I use it for boilerplate that is boring to write and I can check quickly, and to discover things when it's faster to see what copilot does than google around
You'll eventually get the hang of it to know when it's worth using it and when it isn't
-
RE: I, ChatGPT
@izzion said in I, ChatGPT:
@sockpuppet7 said in I, ChatGPT:
@izzion said in I, ChatGPT:
The work of the people making the AI is at least as expensive per unit worked (ok, a lot more expensive per unit worked) than the work the AI is saving on.
That would happen only if the work for running the AI is more expensive after it is created is more expensive, and it stops improving
There is a huge difference between gpt 3.5 and gpt 4. And a huge cost reduction to run gpt-4o compared to gpt-4. Also llama3 api cost 1/100 the price of gpt-4, and is good enough for most of the simpler stuff. That means, we'll either get an AI that solve harder problems, and/or get very cheap to run AIs
I too enjoy getting disbarred for very cheap because I replaced all my paralegals with cheap to run AIs.
That would be the same than if that lawyer hired a cheap untrustworthy contractor to write that for him and didn't check. He ignored the label that say:
-
RE: I, ChatGPT
@izzion said in I, ChatGPT:
The work of the people making the AI is at least as expensive per unit worked (ok, a lot more expensive per unit worked) than the work the AI is saving on.
That would happen only if the work for running the AI is more expensive after it is created is more expensive, and it stops improving
There is a huge difference between gpt 3.5 and gpt 4. And a huge cost reduction to run gpt-4o compared to gpt-4. Also llama3 api cost 1/100 the price of gpt-4, and is good enough for most of the simpler stuff. That means, we'll either get an AI that solve harder problems, and/or get very cheap to run AIs
-
RE: I, ChatGPT
@Arantor said in I, ChatGPT:
Which for far too many people is the new normal.
That is the same old "doing it wrong" but with AI
-
RE: I, ChatGPT
@DogsB said in I, ChatGPT:
@sockpuppet7 said in I, ChatGPT:
@DogsB said in I, ChatGPT:
Generally what I've been thinking.
misleading headline, it say about the work for the people that keep the AI online and etc. for people using it that is still a time saver
Not when you have people double checking its work so that it hasn't made up something again.
"When"
If you use it to create things that you'll take longer to check than to write yourself you're
-
RE: I, ChatGPT
@DogsB said in I, ChatGPT:
Generally what I've been thinking.
misleading headline, it say about the work for the people that keep the AI online and etc. for people using it that is still a time saver