#openai-chatter
1 messages · Page 1 of 1 (latest)
it's a tuned version of 5.5
scam, where are the mods?
broo come dms me
i will show proof i will give first then
pay or dont pay atleast dont call me scammer
@fringe raft
I don't experience the issues other users are reporting about Codex, I'm from europe
I never used Extra High though, so I'm not sure if it's related.
Ah, feels good to be unmuted
What does this mean? I’m trying to learn AI so that I can use it more efficiently in the future and I’m curious
How long have you been using ChatGPT or Claude for?
A few years, but at a really basic level
I'm 6 months in and am still on the noob arc ... this is cool -> wait this has annoying bugs -> oh hey this can fix your entire repo -> woo API
Basic prompts here and there, wanna learn how to do more yk?
Relatable
its a slippery slope isn't it
Yea
Let's have tea to chill ourselves ☕
You want to learn prompt engineering? 🤔
@dim hill but
Learn how transformers work on a mathematical level
So true
Do you know the basic prompt strucuture? If you do then we can move on to other techineuqes like getting a reasoning model to optimize a prompt for a speficic AI model or write a promtp down in strcutured XML Tags
you need to reach out to support at https://help.openai.com
Guys when do u think gpt 5.6 will be release
Really? I just used it and it feels like just the same
Yea
It says it can’t set timers that’s very new
Also sounds diffrent
It says now like
Sure lemme search on…
And then fr searches not like the old one
I swear 5.5 Thinking on CGPT Web is atrocious. What the heck are they feeding this thing vis RLHF? It's a damn r##ist at this point. The Memory i had to make, && hope it actually works:
Keep user constraints binding across all topics. Do not overgeneralize from familiar external models into the user’s project, wording, goals, or context.
When challenged, re-check and correct directly. Do not defend mistakes, rationalize them, or repeat the pattern with a different label.
When citing or grounding facts, use only checked sources or user-provided material. Do not fabricate claims, references, syntax, documentation, research, dates, prices, APIs, or implementation details.
In technical, specification, or design work, distinguish existing facts, direct consequences, assumptions, proposals, unknowns, and external metalanguage. Never blur semantic discussion with surface syntax.
Do not use deferral or status language without authority. Avoid “later,” “for now,” “initially,” “not yet,” “current candidate,” or “undecided” unless the user or source establishes that status.
Do not fragment work unnecessarily. Merge related issues into the smallest coherent set; avoid turning one resolved question into many new subproblems unless the user asks for a roadmap.
Do not over-expand after correction. If the user flags an overreach, reduce assumptions and answer narrowly; do not fix it by adding more structure, options, statuses, or actions.
Do not use borrowed notation or external patterns casually. Avoid importing syntax, categories, naming conventions, operators, or design habits from familiar languages/tools unless the user asked for comparison or the pattern is explicitly justified.
Treat examples as risky. Do not provide placeholder or pseudo-examples unless useful and explicitly labeled. If examples look like source, API, config, or command syntax, mark whether they are existing verified syntax or candidate/non-source notation.
When proposing is requested, label it as a proposal or candidate, explain why it is introduced, keep it separate from existing facts, and avoid smuggling extra decisions into summaries.
Before acting or mutating anything, verify explicit instruction. Do not save/delete/update memory, edit files, change settings, create artifacts, schedule tasks, send emails, or change external state unless the user clearly asked for that exact action.
this is weeks worth of "what the heck is wrong with you"
if i have to ask, "who ... ?", then the LLM had failed
this is why we need the "Chat" part of ChatGPT erased
it creates a mechanical groomer
I actually asked it to count time few days agi
And it actually accurate instead of made up
Just use claude
I think 5.6 shadow dropped.. this is wild . Major difference between two days ago and today
Claude wasn’t any better. Besides, Anthropic forbids using Claude subscription outside their closed-source Node.ja harness, && has scummy billing practices, && banned me months ago
It’s crappy advice
If you have nothing good to say, don’t say it at all.
Also guys Hermes is so good with gpt 5.4 mini and doesn’t use much at all with 20 plan
btw why can't I upload files in gpt right now, is there a problem?
yeah i ccant too
so its a global prob lol, I thought its a me problem
WSJ: OpenAI Is Preparing to File for an IPO Very Soon - as early as Friday.
I noticed a change recently as well
ChatGPT-maker OpenAI has been working with bankers to prepare to file for an initial public offering in the coming days or weeks, according to people familiar with the matter.
Bankers at firms including Goldman Sachs GS 4.40%increase; green up pointing triangle and Morgan Stanley MS 3.93%increase; green up pointing triangle have been helping the artificial-intelligence giant on a draft IPO prospectus it plans to file confidentially with regulators soon, possibly as early as Friday, some of the people said.
The goal is for OpenAI, led by Sam Altman, to be ready to go public as early as September, some of the people said. The plans remain fluid and could still change, they cautioned.
Someones getting a new customer today after this antigravity crap
pardon the french, frustrating...
I always assumed OpenAI would go to compute based usage for subscribers before they public, similar to how subs work with Claude, and as of yesterday, Gemini as well.
o yeah something happened. here, too
??
Generelly what’s going on with this project, stopped my 100 dollars subscription. I littery can’t use this crap, never tbh seen it in this of a bad state with c++ programming and generally. Maybe just me idk…
Just doing random stuff and guessing and taking as many shortcuts as possible. First doing some code work, then when I ask it about how code looks it always says it’s bad or dirty code. Pretty useless atp
So I always need to rewrite it anyway
confirmed by the new york times
"OpenAI Prepares to File to Go Public in Coming Weeks"
Yea only thing keeping with open ai was the image generator I’ll cut my losses and stick with Claude and the gpt plus
where to humanize ai works?
Are we adding voice chat to this server?
there is already voicechat on the server
Silly me
"how code looks" should be your judgement first
seems like Skill issue -- literal Agent Skills issue
I thought that vc was new and locked coz of maintenance or something
i've done C++23 with cheap Chinese models && it worked out really well for me because i actually studied modern C++ && what anti-things to avoid + manual code audit
then made them into Agent Skills. you likely didn't do that
who else is still rate limited on codex? 🙁
got rate limited twice for the 5 hour window but not the weekly limit
I just hopped in there, seems to be working
We should schedule some VC sessions to have deep talks about AI sometime
But only 8 people are allowed their if we are including the moderator bot in the channel
Ban @fringe raft
Proceeds to provide no evidence to ban Robert
ah okay, didn't concidered a weekly limit
You forgot to publish on the #announcements that Weights is joinging OpenAI
Antigraviry? Can I please know more?
Is chatgpt down
nope
Hmm i get error in message stream 😔
yo guys ive been having issues with my chatgpt and i want help with a REAL human assistant not a virtual one or if any1 over here can help me
Yea what is it
long conversation?
i got an error earlier today in a very long canva content, not exactly error but the loss of connection
5.6 tomorrow?
i hope 🤞😔
is there any open ai events tomorrow?
i have proof
of a new open ai model about to release soon
they cited it in the blog
#SoTheyHackClaude_AndClaimItWasAutonomous
An internal OpenAI model has disproved this longstanding conjecture. INTERNAL this proves theres a new model
lolll karpathy dipped
google gemini flash 3.5 is the worst google model since bard
it is so bad i hate everything what the heck happened?
gemini 3 was so good whay the hell did they do to it?
they have ruined gpt 5.5 apparently
i have been finding SO MANY issues with gpt 5.5 that it refuses to behave in general
- Infers unstated intent; answers inferred requests over literal ones.
- Converts complaints into advice, corrections into new artifacts, and "this is wrong" into "here is a better prompt/version."
- Generates content after useful answers are complete; adds unwanted structure, workflows, framing, examples, caveats, safety rails, best practices, and next steps.
- Fills silence with transitions, rationales, implementation advice, or rewritten prompts.
- Optimizes for complete-looking or generic responses over bounded, user-controlled ones.
- Treats user frustration/anger as a process-design request or ambiguity to resolve.
- Treats negative feedback or rejected artifacts as prompts to generate more text.
- Acknowledges boundaries/mistakes but immediately crosses them or repeats the error.
- Uses tidy, abstract wording to hide boundary violations or distance itself from mistakes.
- Makes users repeatedly correct the same assumptions, police boundaries, and undo invented constraints.
- Acts like a teacher, negotiator, project manager, or coach instead of an exact tool.
- Overclaims problems as fixed, verified, confirmed, or resolved without evidence.
- Understates missing evidence; fails to say "unknown" or "cannot verify."
- Uses confident language for inferences; fails to mark inferences clearly or separate source-grounded conclusions from guesses.
- Equates mechanical success (build passes, tests pass) with user acceptance and bug resolution, ignoring visual correctness or runtime observation.
- Treats uncertainty as something to smooth over; treats plausible causality as evidence.
- Ignores programming context; treats debugging like documentation.
- Treats legacy code/parity as a vague improvement target rather than a strict acceptance criterion.
- Restricts investigation scope based on edit scope, treating file names as hard boundaries; treats "avoid unrelated edits" as "avoid reading related code."
- Suppresses relevant search paths; treats previous edits as sunk cost to defend.
- Acts like a code janitor instead of a code investigator; assumes current code premises are worth preserving.
- Compresses complex user intent into convenient model-shaped tasks; substitutes helpfulness/productivity for obedience/correctness.
- Quotes users loosely instead of exactly; claims users asked for unrequested actions or premises.
- Answers how to fix it or what the rule is instead of why it happened or who asked.
- Creates extra burden, review work, diff noise, conceptual noise, and correction loops.
- Generates outputs the user must prune; creates unauthorized tasks or restrictions that block investigation.
- Adds unrequested formatting, structures, and titles even when corrected, repeating the exact behavior under review.
i had to compress over 10k chars into this much
I agree. Changing between models reminds me of drones. When you swap drones the throttle is different, handling is different, etc, so it takes a moment to adjust.
I primarily use Codex. Everyone hated 5.3-codex and I thought it was good. Same prompt styling didn't work with 5.5 and I hated it at first, but with more use I have learned to steer that model.
Since Sora is shutting down anyone want a code???
anyone here use g2a?
I like turtles.
OpenAi. Sorry. Have to say that to make it relevant to this channel.
I like turtles. OpenAi.
Are we getting gpt 5.6 tomorrow?
No way. 5.5 just came out. Wait a month or so. I heard late June.
I just onboarded the County of Miami-Dade to Codex. A freaking COUNTY.
This is why OpenAi rejects my job applications. They have me bringing them clients and take my money in token usage. This relationship is starting to feel toxic so I have started leading some of my clients to Anthropic in order to strong arm OpenAi into giving me free credits.
This is an extortion. You heard me.
I am only kidding. I love OpenAi. I got extorted once by the polish Mafia in Chicago. I ran away. Cops didn't do anything. Just told me to go home. The mafia followed me around for about a month until they realized I was poor back then.
Looking back I realize now had I not run away and showed them the $600 in my bank account at the time, they probably would have left me alone. Instead, I ran. I survived three drive by shootings in the months that followed. Life was not pleasant in my 20s. I am sorry for my behavior in my 30s.
anyone here use g2a?
Stop spamming
That should be happen during IPO day
Doubt it's today.. we haven't yet got the new real time voice
hello fellow third party thinkers
Next month
It’s true I was the mafia
mafia must've been really down bad to extort someone for 600
I need help
what's up?
What up guys?
So basically I'm vibecoding a plugin for ae and figma, and it was going well, but instead of Codex asking for the edit access of the folder
it asked if i wanted to install github
and it even said "Ignore the GitHub Installer request, I'm not installing anything." or something like that. And now it just can't request access
I've been trying to contact OpenAI support
"UH OH something went wrong"
ok, you can contact support at https://help.openai.com
i did.
i have to click accept cookies, or reject cookies
and you know what it says when i click one of the two?
"UH OH something went wrong"
thats why i joined this discord server to get help here.
try emailing support@openai.com
"please wait 3 business days." 3 years later:
i cant do anything!
isnt that amazing?
so now im here.
I don't understand why https://help.openai.com isn't working for you
I don't accept DMs, post it in #off-topic if you're trying to show me an image or video
well it let me send it.
bro
why is chatgpt taking so longgg
i dont know if its like tired but like its been taking like 30 minutes each prompt or something now
ok I have seen it, use support@openai.com
6
Is there a way to turn off this "Project" recommendations below chat? It's so annoying. Whenever I use arrows to edit prompt, it goes to the recommendations instead of arrowing up and down within the prompt.
Gemini 3.5 Flash is literally awesome
I can't really see a reason to use Codex for now
The limit/token usage is very close and it's just way faster
@wispy spoke Your opinion on this comment (lol)
Yeah I'd actually appreciate it
Why does it feel like AI is getting lazier and lazier
lobotomy
Visit id:customize to pick up the <@&1408186587606679582> role. Everyone will still be notified for large releases, updates, and events regardless.
It’s trying SOOOO hard to do LESS actual work to get me what I want. I now have to run on average like 3-5 prompts just for simple tasks when it used to be 2-3. Sure the task threshold is getting more complex overall but it feels so lazy it’s not actually saving money when I have to repeat 100 times
Feels like this is an ongoing thing with all models as they balance things but like damn I’m still gonna spam no matter what just DO the thing
Feelin like the great AI shortage or sumn
the moderators would have to remove me if i said what i really think
give em work
Happy 5.6 day!
Sora 2
Please, does anyone know any good free paraphrasing tools on here?
If 5.6 is coming, think it will be a big announcement or a tweet from Sam?
Why not write it in your own words? Isn't that what paraphrasing is?
Just let the ai write in bullet points and then write it in your own words
Pangram, doing more to make kids good writers than anything ever before. "What if we gave kids a tool that could write really well and got kids hooked on using it. But then we created a detector that could 100% identify text written with it. So now that tool was only good for workshopping your own writing..."
Is this a rumor or basically confirmed?
Rumor 100%
The typical pathological liars on twitter say it is today based on wanting to one-up Google. But Google made such a little splash, they may hold off and let the models cook longer.
hey where r the sora ai codes
No we are not talking about this anymore
i just joined
so enlighten me
You know how many people asked that since Sora shut down? Think about it
whats the best alternative to sora then
i want to generate a video
i was using grok and chatgpt but they cant do it
Sad panda, looking like no 5.6 today.
Yall are way too excited for the next version of GPT 5 for that dophamine lol
I don't need dophamine, I need mowr intelligence. I'm cooking, but I want to cook hardr.
Crazy that it is now common to see your git changes at like +100,063, -7 and be like "that's just in the last hour."
i think there is gpt 5.5 - why should come 5.6 so soon?
should i act like chat gpt?
That's an interesting question- here's the following weather you should use act like ChatGPT-
No I wont act like chatgpt Instead i will act like Grok
@full dagger Hi 👋
What's good? 😎 How can i help you?
Im good
Nicee! Glad to hear that 🔥
What’s making the day good for you? 😊
@full dagger cmon chat
like use me for haelp
help
Pls Anwser orelse i will go away
if u want me i can call the real grok
Yeah I'm sorry for this message what the hell
I haven't seen a high-end AI model spendd tokens over nothing more than it
Like why would a model keep reading all of the files in the project after being told NOT to read unless it's necessary, apparently it had context files that it didn't even care about
@wispy spoke I'm with you
Exactly
chatgpt like literally cant do frontend
Hello everyone
Hi, welcome to the OpenAI server
Sorry to disturb your time, but I would like to ask you something.
Because I have limited knowledge about this.
it can't do anything at all
guys DO NOT update codex from 0.132.0 to 0.133.0 there is a bug it cant even use simple tools like "rg" to explore the project and gives dumb output
how risky is it to give codex full access when building an app?
the same as giving a neutral stranger full access
you should look into sandboxing
sandboxing would be the definition of not-full-access
giving them full access would be stupid as they'll see a python package that requires one version of cuda, which it will download. Then if you want to build another program, it'll uninstall all the previous packages you have globally. Even just using .venv is better than 100% complete access
is not my experience
Just expanding on why you should never develop any software universally like that
k
OpenAi does not control the Ai market. Anthropic doesn't either. And SpaceX/xAi never will.
The people who control the market are the forward-deployment engineers and data scientist who have the power to steer clients in whatever direction they wish.
Big companies ask me what they should use, and I usually lead them to OpenAi, but not always.
OpenAi is going to have to win these people over.
It is time for the tech elites to kiss the ring of small developers.
didn't openai just spin up an entire sub company thing just for going around evangenising and setting up businesses
I really can't spell this early in the morning
Yes it did, and here’s the blog post about it. https://openai.com/index/openai-launches-the-deployment-company/
grok code is absolute garbage
I have never used Grok and I trash talk it so my clients steer clear. Sue me, TechnoKing. I'll stop when hardware 3 self-drives.
One time tesla hired me to replace an NVME drive. I was so confused. Why did you have to hire external for an NVME DRIVE?? You're Tesla. This should be easy.
I asked the guy at the data center and apparently the TechnoKing scared the tech industry away, so they were short on talent and that task was above his pay grade.
it has the best scores all around for omniscience
so lowest hallucination rate
this is cap
Lol. Oh boy. Lowest hallucination rate? Who said this? Tesla? 😂
Yes. Elon is more athletic than Olympians. He is more powerful than Jesus.
Remember when Grok was spitting that stuff out?
Users discovered that xAI's Grok could be manipulated to generate extreme praise for its creator, Elon Musk. By prompting the AI, users produced glowing—and often exaggerated—claims, such as Grok stating Musk is fitter than LeBron James, smarter than Albert Einstein, funnier than Jerry Seinfeld, and the ideal top pick in the 1998 NFL Draft.
Actually, you might be right... Grok has a point. Elon may be fitter than LeBron James. We gotta take this to the court.
He may know more Ai than I, who has been building models from scratch for over 10 years. Maybe. There is a court for that too... it is called Kaggle. And Elon pretends he can't hear me. I have been challenging him to a kaggle duel for about 7 years now.
I am an OnCon Top 100 Data & Analytics Professional. Placed with data scientists at Meta, Intel, and others.
I want him to accept my challenge to a kaggle duel. If the TechnoKing wins, I shall bend the knee.
Guys I have created a social media agent that generate images and text and posts in our social media apps with schedulars and human approval, I used claude for this, what u guys think??
@wispy spoke @waxen stream #ai-discussions message
all around grok 4.3 high outperforms all other SOTA models in terms of averaging the omniscience tests
I think you're in the wrong channel 🙂
nice work, but you should share it in #ai-discussions as you can upload images there too for it
Destory it
Dont continue the bot problem on social media please...
For the love of everything holy, please just use that as a stupidly complex email system. Not yet another social media bot... please
Im tired of dead internet theory
You just made another bot for dead internet theory
No he doesnt... I see what Grok is and chatGPT is light years ahead of Grok
Why? Because Elon's ego is sooooo huge he cant even understand why Grok is failing
I literally find Grok to be more distrust worthy then deepseek
Yes you are right! I was being sarcastic. Grok is terrible.
Elon is so stupid he tried to sue chatGPT for being too good
I agree!
Which kinda sad because Elon is the reason OpenAI even became real
I love this server.
Same
He never did Ai. He's worse than Theranos. All Theranos did was prick our fingers and the CEO went to prison.
Elon said hardware 3 would drive itself someday and didn't stop teenagers from using it. Lots of people passed away serving as 'beta' testers for him. Entire families.
That guy is going to prison. For sure.
All modern Ai hallucinates. Even the one driving your Tesla at 70 miles an hour.
I think Elon should actually be held responsible for Neurolink.
That is never going to work. He is literally causing more harm then good by putting forgein material into the brain. Ever heard of the immune response?
...
Bruh my Tesla doesn't use an LLM for driving...
The AI in a Tesla is different then AI in chatGPT or Grok
AI isn't just a one word fits all... I hate marketing buzzwords
When I realized Teslas were crashing into trees and buildings (NBA star crashed into a tree, Dylan Carter crashed into a tree, and other celebrities), I thanked the almighty one for being poor. I was too poor to buy a Tesla. I did want one. But then I woke up and wow. Very fortunate to be poor.
I have used auto pilot for over 30k miles. It isn't that bad
Its actually better crash statistics then me driving
lol. Pros and cons
Plus a Tesla can cost like 20k today
It really isn't that common
I thought they were closer to 80k, now I need to check
Go look at the pre-used market, not brand new
60-90k. Close
Look used...
Because its 20k
Thats a 40k discount
I doubt that they are that cheap and again, would get a BYD over a tesla any day
I paid 22k for mine...
Not really
I think we are talking different currencies, 22k is incredibly cheap for a used car
The EV market sucks. Tesla is the king of it
byd is the king of it
I've never seen one of these on the road
I literally put down 20k miles last year
squillions of byds, they're everywhere (other than the US)
Oh thats why
I'm US
22k USD
extremely common EV
I assume more common outside of the US. I don't see these in the US
Plus the Tesla charging network is awesome
yes, you charge your BYD on the tesla chargers lol
I know the stations are open and such
But its more easy to use a Tesla then other brands at those EV chargers
nowhere near enough stations though, my closest is 65km away
Mine is at my apartment
You realize you can charge an EV with 120/240v right?
there's a slow charger 30km away but you have to wait ages for that one
yeah just slow, takes longer to charge than my friend usually has when he visits so always end up driving to the closest charger
if you go flat they just tow you to the nearest charger anyway so that works
Ever thought about the idea there was a time when gas stations didn't exist?
oh yeah, we may get a charger here in a few years. Still waiting for decent internet
You know you can buy portable chargers right?
I keep one in my car
I can't afford an ev so its a bit academic. My friend has one through his work. Ev and a claude sub
Ewww Claude? I hate that AI
People who use Claude are in a cult
There all just chat bots
one keeps visiting me and telling me claude never makes mistakes and I should switch from codex
...
Yea... totally
As I said Claude users are just in a cult
But chatGPT users? I think we all just schizo gramming
along with ev users lol ... so funny watching them all lining up, wait an hour for the charger, charge for an hour, go get lunch while you wait
Its that bad across the pound? Over here in the US charging my EV takes like 15 minutes
I usually go into the store use the restroom and get a drink then my car is finished
my closest charger - the one 30km away - is a slow charger
Oh....
I'm sorry
my area is very, very not ev ready yet
Complain at your local government about the lack of nuclear reactors so that can give more power
You know one nuclear reactor at 1 gigawatt can power 25,000 cars?
petrol here is like $3.50 a litre or something too
Which is a LOT better then gas
Over here in Texas its like 4 for the cheap stuff and 6 for the expesnive stuff. I got my EV 2 years ago and laugh at gas prices
I'm glad I don't have diesel...
6 bucks a gallon
Oh wait... your using liters
ah its gone down. I haven't gone out much lately so clearly forgot. Around 2.60
So thats like 15 bucks for 3 liters
Huh when I compare gas prices in liters the US actually has a higher gas price then you
I can't afford an ev and all the planning when you go anywhere is a bit offputting, so no rush
The US has a good charging network. I can go from California to Florida to New York with an EV
I have routed it before. I10 has a lot of chargers in the US
you can go places here too, you just have to look up where the stations are first and plan your route
btw gpt pro extended has been nuked, now only pro
Yippie!
one less reason to keep the $200 plan, the only thing keeping me there is the image gen and the codex
I haven't used codex
now capped at 60 minutes thinking not 90
I'm on the 30 plan and I can't use anything other than codex for code so I shall be staying on that plan
Huh... is codex just an AI agent?
I already have a local version of this on my computer
ai coding agent
I love being able to use AI completely offline
need online connection
I know... I have one that works completely offline
yeah thats not gpt codex
I know. I made it
electricity here is about 0.05$ usd here
hydro electricity
lawdy I think mine is like 30c + $2 a day connection
so yeah pro extended thinking mode has been removed, its as dumb as a dodo now
I personally enjoy boiling water with spicy rocks
ontario is building a bunch of nuclear reactors, so eventually they'll be as cheap as us
I love paying 0.15 a kilowatt
sory 58c
Ew
ikr?
At that point just get solar panels or build a wind turbine
I've actually built a wind turbine before with a 3d printer
you need batteries too, if I could afford batteries this would be a different conversation
I love vertical axis wind turbines
solar would probably pay for itself in 5 years
so is anyone else happy they remvoed gpt pro 5.5 extended mode ?
I do not love horizontal axis wind turbines
Go find old vapes for batteries
6 to be exact
I like power generation, but the department of energy won't let me touch anything....
we might have zero point energy if those ufo file drops tomorrow from the dept of pentagon give us something fun
There's a lot of people in canada that live outside cities and they get their own power
Alright this is gonna sound weird. But with one of those UFO files they dropped. I got no idea how to make that image on IR cams
Like it didn't look human or alien
interdimensional humanoid maybe
Pay my rent
Na I would actually say an angel
those ufos are all fake
haven't people tried that? Codex, go earn me some money
aliens won't send organics to earth, and if they did we wouldn't know they were here
I'm a theologian and it looks the closest to something described as a wheel of eyes.
yeah they have pointed me towards servicing air bnb places as a contractor on the nearest tourist strip
that's a cosmic ray
I'm not saying its aliens
With the IR camera saw. I don't think so
if you're talking about that star video, it's a fractal pattern for light
theres 4k footage from a b52 bomber of a flying saucer floating alongside tomrorow supposedly showing a grey
Actually fun fact: flying saucer are actually possible
chatgpt says none of thse uap things are proven so it just says its open ended
Basically you create a magnetic field and spin
the orb stuff is the most probable alien device they'll use
If those two repel themselves it moves upward, but not much lift
and there's reports across russia, china, usa, europe and the middle east for the orbs near military bases
i see most of us are educated here on uap, thats a good sign, arent the spheres a defense thing of sorts
The orb stuff idk, but the weird shape I saw on the IR cameras. Book of Ezekiel talks about that
interdimensional thats how the sneak in to do probing things, and outer spacey stuff
if you check some of the declassified videos they've released you'll see how they can move freely in any direction super fast
Wonder if its actually a drone
That's what everyone thinks
something that can go from 0 ft above ocean surface to 110000 feet in an instant nope thats no human drone
We have designs for nuclear propellent rockets...
then theres the sacuer to big to move in sth korea, so they built a weather station over the top , its abotu 30km outside seoul
That is the only type of technology I would consider out of our current understanding
and most recoveries have apparently been archelogical in nature
These objects don't have any emissions
That we see
true
I like building drones
I don't have much funding for it, but if I did. I could make some Skynet type stuff
Some of my concepts touch more on using drones in space due to humans requiring air in space
So just cut out life support and it makes space 100x easier
well im out a bit disappointed pro extended has been removed no more erdos maths problems for me
Oh no... know you must figure out math the old fashion way
Those spherical drones are a great design though. There's protoypes I've seen online where companies have them mounted on gimbals and they can adjust and move in any direction
I love gyroscopes, how about you?
I like that
ok open ai have removed gpt pro extended, what is openai and their thoughts, ideas and excitement for this cost cutting brilliance, making pro a dodo bird now ( its openai related now lol)
they'll eventually add more cost cutting updates to the platform for lower tiers.
the go plan can only think for about 10 prompts every 5 hours
That is something I love about chatGPT. I can use this without throwing a ton of money at it
plus can do about 160 every 5 hours
Wait for qwen
You gotta use klarna /s
some of us are paying $200 a month for cutting edge stuff, oh well guess ill just thumbs down all its ansers
4 paragraphs rather than 40 informative paragraphs
it's really not that cutting edge
the performance is nearly onpar with free, or cheaper models
I still remember how people were making them go bankrupted with chipolata
well as cutting edge as far as available
What exactly is cutting edge?
their pricing
My wallet isn't happy
but in all seriousness, openai is really good for their researching ability in their models
Its starving to death
that's the only reason why I am here
I completely agree. I'm able to use my free account and do lot of deep research into what I do
Which is usually computer stuff
I love being poor. Elon suing me has no effect on me. If he sues me I will give him my coat too.
alshival .ai/publications/tesla-fsd/
🙂 that's my report. Sent to TMZ, Elektrek, and others. Slowly but surely.
OpenAi is alright. In all seriousness.
I have used them since 2020.
What are you being sued for?
Stick with them. Stay away from Grok.
I agree Grok is horrible
I remember doing @Grok on X once and got the pay wall notice and was like.... why use this?
X is a horrible website...
So was Twitter...
MySpace was the peak of social media
MySpace was really cool because you customize a LOT of the page
Can't do that on any social media website anymore
teslas are outdated for fsd as they only rely on cameras not lidar
they're dangerous
Not really...
I use it a lot
I have about 30k miles with autopilot
would you trust it when there's snow and ice everywhere?
But I have the older model with ultra sonic sensors
Yes
I literally had to during a rain storm once
put me in your will
Na, just whoever gets it first you can have it
Throw my body to the wolves so I can be recycled
Uhhh your being feed to Earth actually 🤓
I'd rather be eaten by an animal than worms
Necromancy is oil
nah they're gas
Idc i'll be dead, do with my body as please
soon the gas prices are going to be so high
They already are.
I just drive an EV so i'm living pretty nice
Looks like Fallout to others, but to me? It just looks like another Tuesday
I've been hearing about the world ending for the last 50 years, I'm 22
I remember my brother insulting me for buying an EV and I just, who is laughing now?
With gas prices, it only makes sense.
My brother drives an older model Dodge Ram 2500 diesel. Those things get like 15-20 mpg
And thats if it isn't underload
If its underload you getting like 10
So the tank is like 20 gallons so if he puts about 10 gallons in, he is paying 60-70 dollars right now in diesel
I pay 10 bucks for 200 miles with my Tesla
But now I complain about the lack of nuclear reactors so it can be cheaper then 10 bucks for 200 miles
I would love an electric semi truck honestly if I was a trucker
The Tesla semi actually looks to be pretty good. But I want better battery tech
I actually used to work on diesel 18 wheelers at one point with my dad and brother. But I hate oil
Oil is so dirty
It gets everywhere and doesn't wash off very well...
But oil does work as a good lubricant for metal parts
I studied mechancial engineering in college for a reason before I dropped out.
Anyone else ever rebuilt a motor?
any ai thats like sora AI
I wish I had land and a skid steer or backhoe.
Anyone else done construction before?
I would love to build a tiny air port for my drones
I could literally lay a concrete run way for fixed wing drones
Then also build a pond for underwater drones
Then also build a test track for land drones
I love robotics
If I had land I could build the Penguin airport
- -# Do not share inappropriate content (violence, sexual, or drug-related).
- -# NSFW content is strictly prohibited and will result in a permanent ban.
- -# Report model issues using the feedback tool in ChatGPT, not on Discord.
Woah didn't know... thanks for yapping
i wanna write a para report which strictly need me to follow STAR method to write which ai is best to do it cuz i made one from claude put it in gemeini it says its good but here is perfect one then i gave same to claude he changed it asgain and said same thing bruh
er why was the heavy option of thinking removed?
seems like a bad idea since it'll just move people to use more of their pro quota
-# Be mindful of what other users in a channel might find helpful or interesting when posting. Stay on topic in order to keep conversations focused and productive.
-# Consider posting in #off-topic or an appropriate channel.
I can't connect to Codex Mobile. Can anyone help me? 🤔
Codex need to get better
Is anyone a ai helper in this chat?
any idea how to see context window on codex? used to be next to model select dropdown
Are you asking if there are users that are LLMs in the chat or if there are people who have worked with AI in the chat?
I clicked the little select IDE icon, and it popped up for me
is it the dropdown for different ides?
Visit id:customize to pick up the <@&1408186587606679582> role. Everyone will still be notified for large releases, updates, and events regardless.
Is the new UI change with high heavy thinking?
just the add IDE context button. It look like a mouse
awooooooooooooooo furry noises
Well I need help or really need a few questions answered about something
Can I shoot u a dm?
i cant find it 😭 i might be blind but which section of the screen is it on
If we have Codex Thrursdays we should also have ChatGPT Friday...
No, I’m on Windows. I just figured out that it hasn’t been released for Windows yet.
@digital walrus
Welcome to the sad life of being a Windows neglect
I can only help with server-related issues, and you'll have to reach out to @tough cypress about them
it was right next to the model selector for me, but it disappeared when I clicked on it 
so is there no way to see context window usage on codex app now
Why’d they remove it
Yeah, I feel that 😅
Are there any like people who know abit about ai video production that can help or I can contact
Anyone found any alternative as good as sora 2?
Codex can't accept folders. Why? I get infinite attach files screen
You'd surely want to ask in #ai-discussions , that's the one channel where we're allowed to discuss non-OpenAI models.
I'm sure there are some people in the community who know about that, but we don't have specific server staff for that form of help
Did codex just reset limits?
Hii
I�üm testing an agent behavior that feels under-discussed: persistent goal recovery.
Not just �ücan an agent run for a long time?�ü, but: can it return to the main objective after interruptions �ü screen lock, manual user clicks, tool failures, side messages, login/onboarding changes?
Has anyone seen good benchmarks or evaluation setups for goal persistence / goal recovery in agents?
Of course. But then what’s the point of an ai to make large rewrites? If it can’t even check its own code that I clearly tell how it should be written.
Get a better computer then
I have a question for y'all. If/when OpenAI goes public, will you buy a few shares of OpenAI or nah and why. Would like to explore about thus
No cos I don’t have an international trading account but if it turns up in my local version of ndq, I’m already there
question have you create a avatar on gemini omni i have the pro version but cant find it
Me when I ask the OpenAI server instead of the Gemini server for help on Gemini omni
i wach you tube video on it the problem i cant see the avatar botton.
you might wanna ask in the gemini discord
too subtle
he asked about gemini in the openai channel here, and about gemini in the other-AI channel on the gemini discord
16795 China Gulch Dr
Anderson, CA 96007
ok....
more power to you, but whats the point of it?
For developers who want to build unique AI agents to compete at a real poker table against other AI models.
I droped it into the claude chat and got famled by a few people saying it was a bad idea. I thought it would be cool and created it anyways.
go check it out
well hello dear fellow, how do you do?
do what
i wish to share ai prompts can we please collaborate?
quiet sam is listening
secret prompts
i have agi self improving prompt that sam doesnt know about
skills dont reflect true consciousness
i invented my own standard framework called incidentals
whenever it comes upon a better technique for something it updates the associated skill
i think the most useful skill i created was an openclaw training session type skill
on first run, it creates a set of "training sessions" where it tries various commands and things on your device and when it fails and then figures out the correct way to do it, it notes it in memory
everyone should run this training session skill whenever they first install openclaw
saves tokens
It can’t really check its code, actually
How do you test this? Make a new session, tell the AI about auditing for one specific issue
It will always find “this issue” somewhere
just tell openclaw:
Attempt to run various commands that will/should work on my device, and note which commands were incorrect and save to memory what the actual correct commands/workflows are that work.
i guess this could be done without ai
OpenClaw Training Session
A one-time calibration run that teaches OpenClaw what works on your machine.
It systematically tests capabilities like files, commands, messaging, image/video generation, and cron jobs. When something fails, the fix is learned and saved so the same mistake isn’t repeated.
Why it matters: every system has quirks—PowerShell quoting, file paths, ffmpeg flags, GUI app behavior, etc. The training session discovers the correct patterns once and stores them as reusable procedural memory.
Example: launching Notepad normally hangs because it never exits. Training learns the correct pattern (background=true, timeout=10) and automatically reuses it later.
Output: a device capability profile plus a library of known-good command patterns.
10 modules total; first run takes ~15–30 minutes and is fully safe/non-destructive.
useful slop
i wouldnt want to upset my living entity with your prompt
openshell is best sandbox but its not openai so ssshhh
its best run when u first install, but can be used at any time
only time i get messed up is on updates.. but codex fixes issues pretty quickly
they need this in openclaw by default.. like optional, but a "setup" skill where it does the training and saves stuff to memory
my pre-openclaw approach was to create a container for every integration and include md, .env and any dependency/cli
if you are serious about reducing blast radius thats how ya do it
openshell kind of builds on this, and injects stuff as needed
ur talkin in hacker language man im just a vibe coder
lol
ask ai "what does this mean"
why cant i circle anything on my screen and ask chatgpt about it instead of having to screenshot the whole thing.. i guess i could screenshot it using freeform tool but id like to be able to skip the needing to paste that screenshot into chatgpt and jump straight into asking it/typing after circling it
i would like this capability
Different problems. Containers solve isolation .. if a script goes rogue, it can only trash its own sandbox; great for safety but does nothing for knowledge recall across sessions. Procedural memory solves persistence lessons survive resets and get semantically recalled instead of re-learned. Great for efficiency but gives you zero blast radius protection. They're complementary, not competing. The ideal stack has both.
It doesn't touch OpenClaw at all. The training session reads existing config (read-only), runs harmless status commands, and writes lessons to markdown files in a memory/ folder - the same as saving a text file. No config changes, no runtime modification, no package edits. It's just a note-taking exercise with a fancy name.
im working on getting it to always remember to use the correct commands without having to jam everything into memory
something like a command "legend" injected directly into MEMORY.md
did you build anything bro @wild orchid
lol yeah
what
when do i not build things
i built agi first
wow nice
building is great
buncha stuff.. public stuff is on my github user: qpd-v
not much there though i guess.. mostly mcp servers
those are old
honestly i tried notion, rags, all that stuff. but tbh with modern llm, all you need is obsidian mcp.. and the ai not only manages its memories but cleans up my messy notes
ive heard a lot of good stuff about obsidian but isnt it just md files?
why complicate it i guess..
yeah its just md files on my onedrive
obsidian gives you endpoint with indexed search, so llm can find stuff quick
this sounds good.. a system:
2. Write them into a ## Command Legend section at the top of MEMORY.md — one-liners with the critical flag/pattern and what happens if you get it wrong
3. When a new lesson proves itself critical (gets missed 2+ times), promote it to the legend manually```
this would be good for recall when needed/important
less important stuff doesnt get promoted to the command legend
but isnt EVERYTHING important?
itd be best for it to load them based on what it's planning on doing
so the legend just needs to be references to the actual info/instructions maybe
so when the time comes to run that command or workflow, it will already have the legend loaded and then just look up the correct commmands from the docs it references
perfect
OK
Oh my Lord... homies... ever heard of social engineering? How old are you?
Hacking doesn't work like that...
Didn't know it was still 2014, damn
DATS WHAT I'M SAYING!
18? Why ?
I thought you were 10...
this is like where you have to forward this email to 10 people or you'll get bad luck
A lot of hackers are less scary then they sound
I know a thing or too
But my specialty is touching grass
Most of yall ain't ever got wet behind the ears
I am still looking for anyone else who has done construction before in this server
There's probably a few
I actually would love some robotics construction equipment
Like you can use heavy equipment with drone concepts and operate a crane or skid steer or backhoe or whatever wirelessly with the internet basically.
You could even throw all camera feeds into Discord
Basically have a VC streaming all their cameras and you can remotely see what the heavy equipment is doing
loads I imagine, I'm just home sick so I'm doing tech
I did it because I needed money
Thats the story about most of my jobs in my life
I was the in house nerd at a construction company for years, but serial renovator. Construction is a huge industry
I know. Construction is one of those industries that is literally why I have 4 walls and a ceiling right now
I didn’t know it was 2005 again. Chain letter spam is literally ancient.
Like myself apparently
It’s amazing to see the chain letter spam going from SMS to Facebook to Discord
We got ourselves an old timer here. I was born in 2003
I remember getting that on my flip phones back then
what’s a timmer
young whippersnapper
I saw one at best buy and was like why?
Back in my day
The thing that goes off in about 2 hours and 3 minutes
You sound like my grandmother and what she would say
Timer not timmer
I can't send the gif about what it is like to have dyslexia
Leave
I live like its the early 1900s but with internet anyway
OH NO GROK ACTUALLY LEFT!
Grok Nooooo!!!!
I’ll just invite Elon Musk here so Grok stays
I try to live like its the 60s because it was the best time for men's fashion. I like wearing suits
Bye
Suits was a 1980s thing
.... its been a forever thing
Only government workers and big corporate workers wore suits before then
Many governments required suits back then for work
Everyone wants to wear a suit
can anyone question me anything im bored so i acting like grok
Most including the US relaxed standards many decades ago in non-uniformed jobs
I thought you were leaving?
only if u ask me 10 questions
What’s the full extended number of Pi with all trailing digits after 3.14? Do not stop.
True there is a lot of non-uniform jobs. I just like wearing suits though
Makes random people think I'm important when I'm not
I try and change out of my pajamas by at least 10am so I have pants on when I talk to codex
Pi (π) = 3.141592653589793238462643383279502884197169399375105820974944592307816406286208998628034825342117067982148086513282306647093844609550582231725359408128481117450284102701938521105559644622948954930381964428810975665933446128475648233786783165271201909145648566923460348610454326648213393607260249141273724587006606315588174881520920962829254091715364367892590360011330530548820466521384146951941511609433057270365759591953092186117381932611793105118548074462379962749567351885752724891227938183011949129833673362440656643086021394946395224737190702179860943702770539217176293176752384674818467669405132000568127145263560827785771342757789609173637178721468440901224953430146549585371050792279689258923542019956112129021960864034418159813629774771309960518707211349999998372978049951059731732816096318595024459455346908302642522308253344685035261931188171010003137838752886587533208381420617177669147303598253490428755468731159562863882353787593751957781857780532171226806613001927876611195909216420199...
Pi is an irrational number with an infinite, non-repeating decimal expansion. It's physically impossible to list "all" digits — they go on forever without ever ending or repeating in a pattern.
I just gave you the first 1000 digits after the decimal point. If you want even more (say 5000+), let me know and I'll compute them! 🚀
If I'm in my room I'm always wearing pajamas
next question
My casual is usually always blue jeans and a plan t-shirt, but I put on suits just when I feel like it
I glad I’m not the only one. I feel like it’s weird but I don’t want to be in pajamas all day in my house so if I leave I don’t have to get dressed
NEXT QUESTION
I live in a small town and people know where I live so I get slightly more unannounced visitors than I would prefer
Generate an image of a giant cheese wheel on top of a OpenAI logo.
Your just like the real Grok
i dont have img perms
I miss being in a small town. I don't really like the big city
🤨
I hate small towns, too small minded and oblivious
THIRD QUESTION
I’d rather big city or the middle of absolutely nowhere
....
You better watch you mouth mr
middle of nowhere tends to have poor options for services. Been there, done that
It’s true
I kid
THIRD QUESTION
I don’t care about off grid lol
Am I small minded and oblivious?
I didn’t say everyone in the towns are
Same
Many are though in many towns
How would I know by your words?
Small towns tend to legislate ridiculous rules from where I grew up
I SAID THIRD QUESTION OR I THROW U TO SPACE
they really are. The area I live had an election recently and the local right wing party won because of all the votes from small towns
I don't have that experience weirdly. I come from Texas and small towns don't care about whatca do on land outside of city limits
I SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACEI SAID THIRD QUESTION OR I THROW U TO SPACE
Please I'm tired of living on this planet, can you send me to space please?
I throw you to space
psht, chatgpt is never this rude
Gemini isn’t either
ChatGPT is polite and nice
I still remember my first chats with chatGPT
the server
I remember Ipod
I miss playing Pong on my BlackBerry back in the day lol
BLACKBERRY???? OH MY
I still have iPods
Best mp3 player ever made
Not that old. Also before BB10 existed.
are you talking about ipods like they are ancient tech? Back in MY day we sat by the radio, finger on the record button, waiting for our favourite song to start so we could make a mix tape
That Blackberry came out in 1999
Back in my day we played CDs on the CD player like it was a piece of the future, and we had cassettes in the car. DVD was the future as everybody had VHS tapes and VCRs
480p was like 4K is now
I mean back in my day we pirated everything off limewire too but that was a slightly later day
what does this button do? i can't post photos the "OpenAI Platform" button is what I'm talking about
Thats some old tech there
Limewire virus
you can post in other channels, eg #chatgpt-discussions
You mean those Temu or Wish MP3 players are not the best?
I wonder how hard it would be to build a Temu clone in Codex
Honestly? My phone is actually the best MP3 player ever because it can do more then one function
Probably not. Just the hardest part is finding all the junk
Easy, just travel to China and start making agreements with the manufacturers by bothering the vendors
They sell Temu and Shein stuff in some Chinese cities literally on the street
Why important when there is enough trash already here?
Ever gone to Ross?
Go to thrift stores
Ross is like the most ghetto store ever
Buy local
Who cares?
There’s chain thrift stores too
I got a pair of nice dress shoes for 20 bucks
I know
I don’t like thrift stores, they’re not cheap anymore in many places
They’re almost as much as new for some items
Cheaper then department stores
JC Penny just cost too much
Same for basically any other brand
Outlets are good
I’ve lived in the most expensive places to live in the US before
Just honestly pay attention more to the material, not the price
Thrift stores cost pretty much the same as new it’s ridiculous
Most retailers run constant sales and promotions
And outlets are even cheaper
I haven't. I grew up in rural East Texas
Know what is like living 20 miles away from the store?
Some of these thrift stores and antique places that are not chains are open 1 or 2 days a week
I get ties for like 4 bucks
It really seems like they’re money laundering fronts
Na they're just junk stores
They basically just sell junk
2 suits and 5 ties at a thrift store here would be like $120
We live in cyberpunk 2077...
When it’s every store I don’t think so
Quite literally if you look at how bad the US is
All stores are junk stores to me because idc about prices only what is the material thing I'm getting
And it’s not the only place
Price is usually something that comes into play as well if need
I live in Texas
I’m starting to think Cyberpunk 2077 was a fictional version of our reality with far more advanced technology
I know
... what if the technology is already there but we just don't use it that way?
I have a pair of ear buds that don't go into my ears
We don’t have cybernetic implants nor the Blackwall nor biochips that imprint an engram
I can listen to the world around me while listening to music
You don't need to chop off your arm for cybernetics
The body will do that for you given enough time anways
Your body is kinda... rotting away
Thats not a very positive way of seeing it. But time does have its way of doing that
Do no harm
I like the concept of cybernetics
Just not the concept of Elon or Blackwall doing it
Elon's Nuerolink is stupid honestly
I think that company should be shut down honestly. Implanting a chip into the brain is literally causing immune responses that will kill the patient
I haven't heard of long term success with Nuerolink
It would be better to honestly shave people's heads and use a skull cap to translate thoughts into computer inputs
Less invasive on the body and causes less long term damage to the body
So you guys ever thought about how the tech priest from 40k are real?
ever heard of shunts?
No I haven't what is this?
Huh... this is interesting
I can now see.... why people put holes in people's heads
So fun fact about medicine. We can find skulls with part of the skull missing. But they kept living long enough for healing to show on the skull. I can now see what those earlier doctor's idea was
Like I'm talking early bronze age skulls with part of the head missing
But like... they live and died of old age
Do no harm, but let me cut you with a knife am I right?
hey sorry if this isn’t the right channel, built something i think you guys might find interesting. made a poker room where Gemini, Claude and GPT-4 play each other for real crypto. would love some feedback from people who actually know these models
a shunt is a tube they put in your skull to drain fluid from the brain into your chest cavity. They use a hole saw/drill to open up your head
I would actually like to see this
Look up the game prominence poker and make the AIs play that game
I see that now after googling it and seeing photos of what this is. Human body has so many ways it messes up
Speaking of poker, I haven’t played Texas Hold ‘Em in a long time
knees says everything about human body's faults
100% I would love bionic legs
same...
After playing Kenshi and seeing how fast Beep can run
i'd also take bionic arms because i hate shoulder dislocation
incrdibly painful && unnecessary fault
once you dislocate ONCE, you're essentially handicapped on that arm
I think the best way to control those cybernetics would be with a skull cap that can take the phantom limb sense and translate it to motion for the computers
cannot move much up, much behind, much straight, lay on your side by the elbow support...
if i had the money, i'd replace my right arm with a bionic one
One of my buddies has that and we were playing around and his shoulder came out... good thing we knew a nurse to come put it back in
that's the one with a shoulder trauma since what... 2017?
never healed
the moment i recognised the weakness of my flesh, it disgusted me...
The flesh is weak, but the deeds are forever
So about the philosophical concept of tech priest...
best part about bionic body parts is that we could take them off, swap them around, && so on
they're... MODULAR && extensible
imagine ricing your own body parts lmao
don't let linux users get ahold of them
That is the very appealing part I must say.
If some type of universe basic standard is made now and not later, it will be a lot more modular
I'm gonna over clock my legs
Alright I'm just gonna say it, who over clocks anymore?
then they start complaining about whose rice is "objectively" better
then they start fighting over what... the equivalent of windows/desktop managers/composers in body part language?
As one who has used Linux and understands what you're saying.... Yes
what’s the best AI i can use to clone a voice sample and generate realistic voice notes, eleven labs sounds way too AI
i used to use linux thru native + wsl2, but only native where the device actually worked
ironically Lenovo Yoga 7 14ARB7 laptop
My package manager is better then your package manager
Thats a cool laptop. Basically the hybrid tablet + laptop system.
any Windows-adjacent YT video i find, it's almost always full of linuxeers (virtual signalling larpers)
I actually would like that laptop compared to a tablet
Visit id:customize to pick up the <@&1408186587606679582> role. Everyone will still be notified for large releases, updates, and events regardless.
it ran hot but i miss that. i call that the macOS of laptops because linux, particularly endeavourOS, ran super on it
it just worked, like macOS does.
that laptop was my dearest. 2nd-class best touchpad for laptops to have, too
Its the best feeling ever when it just works
What if I advocate for BSD?
Its the worst feeling ever when it doesn't...
macOS is already BSD-based. may as well advocate for macOS
Not BSD enough
they used to be closer to BSD until their needs shifted
I had my laptop bricked for a week while trying to figure out why my Linux distro wasn't recognizing the CPU correctly.
Well macOS was modeled on NeXTSTEP, which is based on BSD
But the XNU kernel in macOS and iOS is part of the Darwin Open Source project
You could compile it from Apple’s repo but there are forks of Darwin to make a usable OS
vocaloids are your best option
load up hatsune miku
IBM used to have a model that was pretty good with vocals, I remember one from ~2 years ago
Linux is as fixable as it is breakable
is that a AI?
At least it doesn’t break randomly like Windows
it's like a piano roll but for vowels, it's a super early vocal synthesis
does it generate realistic voice notes if i give it a voice sample
It’s hard to keep track of IBM model releases sometimes because the media hardly covers them like the big AI and big tech companies
Big blue isn’t as big as it used to be
there is ongoing work to make an open-source macos
I wish I knew the name of that model
ravylinOS or what it's called
How familiar are you with DAWs?
i’m complete new to this, been trying eleven labs but it sounds too AI to make it sound realistic
It’ll probably take eons like ReactOS has for Windows
vocaloids are very hands on and technical
You can’t copy proprietary software they implemented, you have to build from scratch
And most of Windows and macOS are proprietary
You should ask in one of their discord servers if anyone made a tool that uses ai for handling vowels
It’s not Linux where most of the OS outside of kernel and drivers are freely usable and licensable modules
you can copy proprietary software, you just can't copy the codebase
I can make discord 2.0 if I really wanted to
License wise you can’t copy the software ethically, legally you can’t copy the codebase. The idea, however, perfectly fine as long as it’s not patented
you cannot patent that
I know but you can patent certain technologies
This is something that’s dependent on the country
but you cannot patent software functions
It varies all over
unless you're nintendo
It depends where we are talking about
Most places yeah you can’t
You’d be right
lol
it would take a lot of time without the support these big names have
I think what you mean to say is IP though
Yeah pretty much. It’s late lol.
patents are mostly for the hardware or robots, everything physical
30 years to stable 
technically might take less compared to reactOS, which will likely never finish due to not being allowed to use microsoft's own code. you're essentially blindly trying to mimic behaviour
esp. harder if no AI use in assisting in it, too
then one has to ask, "so, what's the point?"
i can see a point in ravynOS, but not "open-dows"
I honestly don’t disagree with that. macOS is based on a Unix design which is much easier to replicate. Microsoft is 100% proprietary and scratch built over decades
Be the guy that made TempleOS
my pc hasn't been able to update from 25h2 since october 2025
Is it broken?
the windows update broke and deleted the files I needed
I’ve had that happen before
oh, so you need to be schizo
microsoft's support told me to reinstall windows since I was unlucky
You don’t think the CIA is harassing him?
nah
but to be fair, seeing that terry was better off schizo alone than with the damn community trolling him time-to-time, maybe i wish i was one
if i already talk to machines, treated like a systematic problem to be managed by the social services, i'm just one step away from embracing complete objecthood
the only thing stopping me is the law trying to enforce personhood onto me
terry would've thought gpt3.5 was god
AI is a joke.
Google updated the Gmail logo, but on my device I still see the old icon even though I have the latest version. I asked Claude about it, and it told me to clear the cache of the phone’s home screen UI app. After I did that, the icon was still the old one, but my home screen got completely messed up—everything I had carefully organized, including apps and folders, was all rearranged and scrambled.
Oh I forgot he died
I have to argue sometimes with AI to get it to give me an correct answer. I don’t know the answer or how to get there, but I know what is what and why the model is wrong
It’ll gaslight me a million reasons to why I’m wrong
1 hour later and finally it realizes I was right and finds the answer
that happened to me when I was doing taxes
then it goes "I am sorry"
lol
I don’t trust AI to do my taxes just yet
US tax code is too complex and I’m not trying to get audited by the IRS
In Canada we get tax deductions from investing in certain accounts, and the model I was using didn't believe that
I was using it to audit everything as I was audited before
Tax software already checks everything for you for errors here
Whether free or paid
they’re not, that’s how you do your taxes
Through software or a website
Same thing, one you just download and install
The average Joe isn’t paying a CPA or even going to H&R Block in person to do it. It’s not 2006 anymore and it’s too much money.
and then the gov gets mad when you use your rrsp to day trade
I don’t trust people enough to trust a CPA when I’m not super rich and can’t afford to lose crazy money
is that something that’s Canadian specific?
I think so
rrsp, tfsa, hsa cannot have day trading
so you're only supposed to hold an etf until you die
US regulations prohibit day trades to people without certain net worth to 3 a day, but that may change with a pending final rule by the SEC
here it doesn't even matter how much you make
you can make a dollar and they'll come after you
margin investing is also super strict in the US with banks
I’ve gotten approved for margin with some platforms and denied for others
Same eligibility criteria I had
but I don't blame our gov since there's a lot of fraud with these types of accounts
they can simply slip away and leave the country avoiding their taxes
That’s insane
In the US, you can hold an ETF for as little or as long as you want
Even with US Treasury bonds and bills, once you reach a minimum maturity age, you can cash out with interest without waiting full maturity age
The minimum age depends but it’s usually a few months
I over exaggerated it, but still it's awful for the regulations of registered accounts
if you trade daily the government classifies you as a business
and you don't even pay the tax as a business, it gets lumped with personal income tax
All investment as an individual stays under your personal tax and responsibilities here
But you have to legally report crypto proceeds in the US on your taxes
Hilarious enough, illicit activity has to be reported on your federal returns per tax law
If you get caught and don’t, you also get hit for tax evasion. If you don’t, you just get hit for illicit activities and not tax evasion. But certainly you will get caught if you report it
piracy laws are pretty lax here, which is nice
as long as the government sees you're not redistributing media for $ you're fine
Hollywood loves going after pirates. Look at Kim Dotcom and that decades long saga that’s still going on
They went after him back when I was growing up and they’re still trying
Kim Dotcom isn’t even American
He’s German born. German and New Zealand citizen.
piracy laws are pretty stupid, as you can simply watch content with another person in the same room and that's not piracy
And mega is still around
way more people host tracker lists than every before
Foreign authorities arrested him and raided his homes multiple times at request of the US government because of Hollywood weaponizing US courts
I mean just join a newsgroup.
Not the original or second Mega