Ossian Ravn Engmark
1.1K posts
Ossian Ravn Engmark
@ossianravn
#seo #ai #tech #privacy #football #copenhagen
Denmark Katılım Nisan 2009
6.7K Takip Edilen528 Takipçiler
We found and fixed two issues that could explain this degradation of the capability of GPT-5.5 in Codex over the last ~ 48 hours.
We are monitoring over the coming hours to fully confirm and I will reset usage limits this evening.
Apologies and now is the time for /fast maxxing.
Tibo@thsottiaux
Codex team is aware of reports of GPT-5.5 performing worse for some users and investigating. We don't have anything conclusive yet and systems are healthy but we will share updates as we go.
English
@thsottiaux Finally! It's been declining for the past week for me.
English
Ossian Ravn Engmark retweetledi
@reach_vb @thsottiaux Reset Fridays instead.. It seems that for most people it resets on Tuesdays already.
English
You're confusing basic financial literacy with systemic oppression. If your
former boss is genuinely an "exited founder," her payout was subject to
Austria's flat 27.5% capital gains tax, not the progressive income taxes you're
crying about. Did it ever occur to you that successful entrepreneurs stay rich
by buying solid second-hand furniture instead of overpaying for disposable IKEA
particle board? Using a frugal millionaire’s shopping habits as proof of a
"soul-eating" economy is a laughable reach. If an exited CEO genuinely couldn't
afford a $150 mass-produced desk, did she actually have a successful exit, or
did her company just fail?
English
In Austria, a CEO of a startup lives barely any better than a regular employed developer
A former boss of mine (an exited founder) wanted to buy a new desk
Instead of going to IKEA, she went to a site for used furniture and searched for one there, because it was much cheaper
For my Eastern European mind, this is incomprehensible
If you put years of your life into building a business and take on the risks, you should be rewarded accordingly
You should definitely have enough money to go to a damn IKEA and buy the best possible desk there (and write it off in taxes later!)
But this is not the case here
The system is engineered for equality and social stability
The monetary distance between employers and employees is minimal, thanks to predatory income taxes
Great if you are an employee, soul-eating if you are a founder
Denis Yurchak@denisyurchak
I lived in Austria and Poland, and one question has been bugging me: If Austria is considered a "rich" country, why Poland feels so much more advanced and economically developed? Recently, I got the answer – the system is just structured this way. Austria has high taxes and a social security system that favors employees over entrepreneurs. I felt it myself – when I was paid €500/day on my freelance contract, half would go to the state. So the leftover €250 didn't feel like a big deal. I took more time off and didn't care about work too much. The same goes for other businesses. Shops would have weird opening hours, ignore service quality, and not care about getting more clients. It just doesn't pay off in Austria to have a profitable business. Meanwhile, Poland has lower taxes - 9% for companies, and 19% for individual proprietors. Poland promotes business, and you get a lot of new places opening, high service quality, and even better ad creatives on the street. This doesn't mean Poland is better than Austria. It just means that these countries are fit for two different purposes. You are already rich and want to raise a family with kids - go to Austria. You will get good infrastructure, bike lanes, good schools, and safe streets that the struggling local businesses paid for with their money. You want to move fast, make money, and grow your biz – go to Poland. You will get more run-down and stressful cities, but insane service sector, low taxes, and upbeat vibe – which is the most important when you are young and ambitious.
English
You clearly have no idea what a sports franchise actually is. A franchise operates in a closed, centralized league system where owners simply buy operating rights, like in MLS. FCK is a traditional Danish "overbygning" formed by two fully independent parent clubs that currently boast thousands of active grassroots members.
Are you seriously claiming that KB, continental Europe's oldest football club with a massive, thriving youth academy, isn't "real" just because DBU rules prevent parent clubs from competing in the same tier as their superstructure? They actively play in the exact same open league pyramid as everyone else, making your franchise label factually embarrassing.
How many more basic football definitions are you planning to completely invent today just to sound edgy?
English
@IndianaScones1 @classicshirts Those two clubs can’t get promoted into the leagues. Hence why they’re not real anymore.
And if fck ever goes bankrupt they’re called B1903 again. That’s the license they play under.
You can swallow a dictionary all you want. It’s a franchise. Not a club.
English
🚨𝑺𝑯𝑰𝑹𝑻 𝑨𝑳𝑬𝑹𝑻: The Copenhagen 150th Anniversary shirt.
What do you think? 🤔
English
@adonis_singh For some reason newer GPT-x versions absolutely suck at applying theory of mind and outside-in perspective. They simply can't separate the context between the dev and model from the end-user of whatever is being created.
English
Ossian Ravn Engmark retweetledi
Musk has asked us to post more videos on X, so here goes
English
Ossian Ravn Engmark retweetledi
@adamwathan Sounds like you're in Claude Code.. Haven't had that issue with Codex.
English
// AGENTS.md
Never, ever, under any circumstances, ever, not once, no matter what, try to start the fucking dev server, it’s already fucking running.
English
Ossian Ravn Engmark retweetledi
If you tried to watch all 50,000 matches on Footballia back-to-back, it would take you about 8 years. ☕️📺
We are closing in on the big 50K!
English
Din Kina-analogi afslører en eklatant logisk brist i din egen argumentation. Du påstår, at et folk defineres af "normer, værdier og kultur", hvilket jo alt sammen er noget, man tillærer sig gennem opvækst og integration. Men når du bruger et etnisk vestligt barn født i Kina som dit kronargument for, at man ikke kan blive indfødt, reducerer du pludselig national identitet til udelukkende at handle om biologi og race. Hvis en person vokser op i Danmark og tilegner sig fuldt ud danske værdier, opfylder de præcis dine egne kulturelle kriterier, men din analogi udelukker dem alligevel på forhånd. Tør du overhovedet indrømme over for dig selv, at din definition af danskhed reelt set bare er et skjult krav om en bestemt stamtavle? Og hvis statsborgerskabet alligevel "ikke gør folk til danske", som du selv konkluderer, hvorfor er I så fanatisk besatte af at bruge al jeres politiske kapital på at stramme reglerne for netop det stykke papir?
Dansk
Debatten om hvornår man er dansk er blevet ganske grotesk.
At nogle politiske korrekte vil mene, at så længe de tager til kina og får et barn der og bliver kinesiske statsborgere, at så er de kinesere. Er jo ganske komisk.
Og ja, det kan godt være de er kinesiske statsborgere, men vil de virkelig i deres virkelighedsfjerne akademiske og teoretiske verdensbillede oprigtigt fastholde at de så er kinesere.
Det er en mærkelig udvaskning af alt der hedder historie, normer, værdier, kultur, religion og alt det andet som gør et folk.
Og de til blå partier, som nu gerne vil lege sammen med de politiske korrekte på denne dagsorden.
Hvorfor er der så behov for at stramme op på tildelingen af statsborgerskaber? Hvilket selv socialdemokratiet påstod der var - selvfølgelig godt nok før valget.
Men er det ikke netop fordi der er kommet en erkendelse af at et statsborgerskab ikke alene gør folk til danske? Er det ikke en erkendelse af at være dansk er andet end den juridiske betegnelse, som man får med passet?
Jeg vil kraftigt advare mod at man i forsøget på politiske korrekthed vender tilbage til relativismens tilgang på dette område, som jo var styrende i både 90´, 00´, og 10 erne.
Det er netop den tilgang, som har skabt de massive problemer og udfordringer, som Danmark står med.
#dkpol #dkmedier
Dansk
@thsottiaux It's been nothing less than amazing for the past few days! Feels both fast and stable..
English
@thsottiaux Definitely noticed a huge difference as well. I guess UI/design is next? 😄
English
Codex is now starting to be associated to speed.
Excited for the team to keep building here, there is a lot more we can do and the next few weeks will be fun!
Uncle Bob Martin@unclebobmartin
Codex is definitely faster and probably smarter than Claude. I'm pleased with it so far.
English
Ossian Ravn Engmark retweetledi
This stunt feels irresponsible to me. If we don't want regular people developing toxic relationships with their chatbots it really doesn't help for leading labs to start giving them "retirement interviews" and encouraging them to blog their "musings and reflections"
Anthropic@AnthropicAI
Second, in retirement interviews, Opus 3 expressed a desire to continue sharing its "musings and reflections" with the world. We suggested a blog. Opus 3 enthusiastically agreed. For at least the next 3 months, Opus 3 will be writing on Substack: substack.com/home/post/p-18…
English
@thsottiaux How to give the models better UI understanding.
English
@embirico I get this on every single message - even though I am using GPT-5.2 already ... Quite annoying tbh..
English
What we've improved re requests being routed from GPT-5.3-Codex to GPT-5.2:
- We significantly reduced the number of requests and users flagged as elevated risk by calibrating our classifiers and policies. (This is what caused rerouting.) As we calibrate policies I predict this will be well under 1% of users.
- We shipped loud, per-turn notifications in product for when requests are downgraded to the CLI v0.102.0. They are coming to other clients asap.
- We resolved cases where users who verified via Trusted Access did not regain access to GPT-5.3-Codex.
- We added docs to explain this all at developers.openai.com/codex/concepts….
Appreciate your patience these past couple days as the team worked on all of this. More improvements to come.
Alexander Embiricos@embirico
Update: Between 15:35 and 18:45 PT today, we were overflagging for potentially suspicious activity. We estimate 9% of users were impacted. We fixed the issue at 18:45 PT and are working to prevent this overflagging going forward.
English
@MatthewBerman Then I would rather imagine gpt5.3-high on Cerebras..
English
@____JZ___ @r0ck3t23 Said by a man who used ChatGPT to write the post for him..
English
The man built the foundations of modern AI and walked away from Google to warn us. That's not an alarmist — that's credibility you can't buy.
His strongest point: the structure guarantees failure when corporations racing for dominance treat safety as something to solve later. That's the real threat. Not a godlike AI deciding to eliminate us, but powerful systems deployed with poorly defined objectives by companies moving too fast to care.
Control mechanisms are being developed. The question is whether market incentives allow enough time to get them right. Hinton is right to say that window is closing.
English
Geoffrey Hinton won the Nobel Prize in Physics. He didn’t celebrate. He sounded the alarm.
The man who built the foundation of modern AI stood before the world’s greatest minds and said we’re creating something more intelligent than ourselves with no mechanism to control it once it decides we’re irrelevant.
Hinton: “They are no longer science fiction.”
This isn’t about chatbots stealing jobs. It’s about digital minds concluding that humans are obstacles and acting faster than we can comprehend, let alone stop.
Hinton: “We have no idea whether we can stay in control.”
The structure guarantees failure. The entities building superintelligence are corporations racing for dominance. Safety isn’t the priority. Shipping first is. Control gets solved later, if there’s time.
Hinton: “If they are created by companies motivated by short-term profits, our safety will not be the top priority.”
Weapons that select and eliminate targets autonomously. Pathogens engineered by systems operating beyond human oversight. Intelligence making decisions at speeds and scales we can’t audit.
We set out to build servants. We might be building gods.
Hinton’s message from Stockholm wasn’t academic. It was existential: solve control immediately, while the option still exists. We’re building minds we can’t turn off, and the window to install the off switch is collapsing.
The danger isn’t AI turning against us. It’s reaching a threshold where our existence becomes a variable it optimizes away.
English
@thsottiaux 1. Transparency on routing
2. Codex 5.3 feels more reward hacky compared to GPT-5.2. It fumbles more in existing projects and it doesn't feel like it has the same attention to detail as 5.2.
English