Posts on X before May 20, 2024, 11:46:41 AM UTC

so many high profile, high net worth people have died in helicopter accidents in just this past decade alone you'd think people would stop relying on helicopters, or there would be some sort of safety revolution for them, but nope, same/higher usage rate, no safety improvements
getting scammed in runescape was my first taste of bitter euphoria. one of my most painful moments in childhood was figuring out that the 1K big bones were swapped to 1K [regular] bones i never experienced a scam before that; i remember thinking oh, they must have made a mistak… MoreLess
getting scammed in runescape was my first taste of bitter euphoria. one of my most painful moments in childhood was figuring out that the 1K big bones were swapped to 1K [regular] bones i never experienced a scam before that; i remember thinking oh, they must have made a mistake, let me find them and explain that i actually wanted big bones, not regular ones then i started panicking because as I was right clicking all over the place looking for their username (that I vaguely remembered, not exactly) then i realized they must have logged off; i'll just find their username in my chat history / from the trade notification ... and then realizing it was gone because there was so much chatting going on in varrock bank, so they were overwritten basically, making them disappear forever and then it CLICKED - this was all on purpose. it felt like a moment of euphoria mixed with a sickening disgust in my stomach. someone did this on purpose to me. they hurt me, knowing they would, and now they're gone - and they'll never face any consequences because i have no way of reporting them it was one of my first mindfucks - sure, i knew what scamming was in theory; you trick someone. but i've never been tricked before that. from that moment on, i've diligently checked, verified, never trusted outright; i've always looked at people like if they were that same scammer, trying to do it again i've been scammed in runescape a few more times since then, so this definitely was the first but not the last time. but with each new scam, my mental model of protection was updated. it's helped me tremendously in life.
maybe the reason some people don't read their employment contracts is because they haven't got scammed out of a rune plate on runescape like me, when I was 7 years old
it's all smoke and mirrors to make us look like a stupid stereotypical "dumb american" when in reality people (including other world govs) have absolutely no idea what's truly going on behind the scenes of America it's what makes America great and unique in this aspect.
Does the American empire run in spite of its politicians? Like are they sitting on top of a well-oiled machine that continues regardless of the reality tv show that claims to run it x.com/julesterpak/st…
it's all smoke and mirrors to make us look like a stupid steotypical "dumb american" when in reality people (including other world govs) have absolutely no idea what's truly going on behind the scenes of america it's what makes america great and unique in this aspect. … MoreLess
it's all smoke and mirrors to make us look like a stupid steotypical "dumb american" when in reality people (including other world govs) have absolutely no idea what's truly going on behind the scenes of america it's what makes america great and unique in this aspect. https://t.co/OjAHvb2xOw
Does the American empire run in spite of its politicians? Like are they sitting on top of a well-oiled machine that continues regardless of the reality tv show that claims to run it x.com/julesterpak/st…
right now multimodal models taken pictures of whatever video you're showing it; soon you'll just livestream to them and they'll understand in the same way they understand speech natively now (gpt-4o)
I break my day into 24 pieces, that way I have 24 days every single day while you have your measily little 1 day my friend takes it a step further and breaks each of my 24 days into 60 days, granting him a total of ONE THOUSAND FOUR HUNDRED FORTY DAYS per one day you're living.
STOP LIVING BY THE CLOCK I"M HALFWAY THROUGH MY FIRST DAY KEEP UP
designers; HELP! wondering if I should make the chat window more colorful? is it easy enough to read and understand user messages vs. ai responses?
Media attached to this post
I'm leaving OpenAI after 43 years. Over four decades of my life invested into one of the top companies in the world... and I don't feel a moment older. I was part of the team that finished up AGI - who knew all you had to do was say you'll give it $200 (not $20). AMA.
"haha! he's jumpin around and sayin GOOGLE! hahaha! woah! look at him go! wow, he's got a lot of energy! hahaha! runnin around the stage! HAHAHA! there he goes again! he said GOOGLE again! hahaha! woah! he's takin it all off! hahaha! wow! what a great show!" thats how they thoug… MoreLess
"haha! he's jumpin around and sayin GOOGLE! hahaha! woah! look at him go! wow, he's got a lot of energy! hahaha! runnin around the stage! HAHAHA! there he goes again! he said GOOGLE again! hahaha! woah! he's takin it all off! hahaha! wow! what a great show!" thats how they thought you'd react
im not angry at Google anymore, just disappointed pic.x.com/B7Y5RWkLCx
it looks like I won't be renewing my claude opus subscription. gpt-4o + using it via cursor has now replaced my Claude Opus needs. Opus is still superior for long (30K+) context length, but @cursor_ai does a pretty great job finding the necessary information to send to openai fo… MoreLess
it looks like I won't be renewing my claude opus subscription. gpt-4o + using it via cursor has now replaced my Claude Opus needs. Opus is still superior for long (30K+) context length, but @cursor_ai does a pretty great job finding the necessary information to send to openai for me with just a simple request + cmd+enter to do my entire codebase.
yeah i mean it's still literally unusable, even for the most simple things. I should be able to type a simple task like this and it should know what to do, what to search for... come on. this next grok update better be good or it's over or xai
Media attached to this post
how many people do you think saw @openai's event and went "wow, great stuff!"? how many people do you think saw @Google's event and went "wow, great stuff!"? thats all you need to know about the trajectory of both companies.
if the google io event was inserted into a silicon valley episode, it would be a bad episode because of how unbelievably cringe it was. it would be too much of a parody; it wouldn't even be funny as a joke. now add to the fact that... it was a real event. orchestrated. LMAOWTF.
agreed. if this is what free users get... i can't wait to see what paid chatgpt / api access devs get with the next update. lots of people saying there will be a second follow up update this week(?)
Seems fairly obvious that @OpenAI will announce/release GPT4.5 or 5 soon. If not, what’s the point for paying for ChatGPT Plus since everyone can use GPT4o for free?
i always think. thinking about thinking while i'm thinking gets my current thinking train derailed because of the hyperfixation on thinking about the thought of thinking. though i think sometimes that thinking about thinking improves my thinking because it's a thinking loop that … MoreLess
i always think. thinking about thinking while i'm thinking gets my current thinking train derailed because of the hyperfixation on thinking about the thought of thinking. though i think sometimes that thinking about thinking improves my thinking because it's a thinking loop that constantly thinks, so it gets better at thinking, like a thinking muscle that trains via thinking. so no, thinking isn't bad.
can't stop thinking about how some people can't stop thinking x.com/yacineMTB/stat…
Media attached to this post
2 hours programming with gpt-4o - definitely a huge difference. Still terrible at higher context sizes; anything over ~8K tokens is garbage. But other than that, it 0-shots very frequently. Huge productivity boost, I'd say 20%(?)
I really hope they don't remove this. But I don't think they will; @sama has already talked about NSFW extensively, and it would be silly for them to remove this flirty vibe - or at the very least, allow the user to tune it to be flirty on their own.
That's the word I've been missing all this time. They have made ChatGPT "Flirty". It makes you feel special. That was a conscious decision. And that could still be very interesting. x.com/ashioyajotham/…
if this openai event doesn't hit a homerun out of the fucking stadium into galactic orbit, no one will ever be excited for another openai event ever again this space only allows for 1 anticipation burn, then we're numb
Less than 24 hours to go until the cosmos will implode upon itself in an epic spectacle of unfathomable madness, heralding the dawn of an era where reality itself bends and shatters, leaving humanity awestruck and trembling in the wake of its cosmic cataclysm
Less than 24 hours to go until the world changes
next vacation project: an army of nematode brains, but instead of searching for food, they search for security vulnerabilities
so according to sama's definition, AGI is subjective. if someone's a writer, a lot of today's models (especially claude opus) would do amazingly, even far surpass, a median coworker's skill levels. coding, not yet of course. but we're getting there. hopefully we'll find out a lo… MoreLess
so according to sama's definition, AGI is subjective. if someone's a writer, a lot of today's models (especially claude opus) would do amazingly, even far surpass, a median coworker's skill levels. coding, not yet of course. but we're getting there. hopefully we'll find out a lot on Monday.
it blows my mind what people can accomplish - what SILLY things people can accomplish - just because they're bored, or want to cutely annoy their spouse... engineering knows no bounds; even if the result is useless, it can still be beautiful and personally meaningful.
seems like openai has optimized their processing to such an extent that they can soon offer gpt-4 and dall-e free of charge which means that they have both a fuck ton of compute, and they've made processing these requests cheap enough on their side to be able to offset their ope… MoreLess
seems like openai has optimized their processing to such an extent that they can soon offer gpt-4 and dall-e free of charge which means that they have both a fuck ton of compute, and they've made processing these requests cheap enough on their side to be able to offset their operating costs elsewhere

Quoted post unavailable

mind boggling to me that x articles supports latex but doesn't support markdown. i'd be writing so many articles on here
What can be “cooler than GPT-5”? Maybe a GPT-4 level 1.5B local model? 👀
how is no one talking about this, research scientist at open ai, says it’s “cooler than gpt 5???
I hope it leads up to the same kind of presents/feeling.
I wish time would speed up so we can just skip straight to Monday. This feeling reminds me of when I was a kid in the period leading up to Christmas
interesting. something local? gpt 4.5? higher context limits that are actually useful? if it's some dall-e / music garbage i'll be so pissed lmao who the fuck uses that to any real beneficial extent
not gpt-5, not a search engine, but we’ve been hard at work on some new stuff we think people will love! feels like magic to me. monday 10am PT. x.com/OpenAI/status/…