Flowers from the future
@futuristflower
Followers
14,020
Following
139
Media
85
Statuses
895
In the unity of all, under night's solemn sky, she stands, sacred and forlorn, with a cunty gleam in her eye.
Joined October 2023
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
Chelsea
• 287384 Tweets
#LAMeetandGreetxENGLOT
• 133284 Tweets
Celtics
• 80853 Tweets
Pacers
• 65981 Tweets
Tatum
• 29577 Tweets
Mikael
• 24783 Tweets
Barco
• 22523 Tweets
O Vasco
• 22108 Tweets
Omar Geles
• 21844 Tweets
Gaston
• 20796 Tweets
Fortaleza
• 20354 Tweets
Herrera
• 19331 Tweets
#WWENXT
• 17453 Tweets
#Aashirwad
• 14934 Tweets
Haliburton
• 13335 Tweets
Temperley
• 12572 Tweets
#サブウェイ新サンド
• 11283 Tweets
Tyrese
• 10259 Tweets
Vallenato
• 10060 Tweets
Jアラート
• 10022 Tweets
The ability of GPT-4o to interpret faces is at human level.
126
385
6K
The laughter is so magnificent and seems so genuine. Many people will find it hard not to fall in love with them.
118
179
2K
GPT-4 wording changed from most advanced to just advanced.
37
78
2K
Oh come on. This is so bitter and in bad faith.
You trained Grok on GPT-4 output. You only made it open source after people highlighted your double standards. You literally asked to pause global AI development when you were busy creating a synthetic GPT-4 dataset.
Come on.
52
83
2K
When you have access to AGI and don’t need to deal with chatgpt like us normal dogs:
35
103
1K
If they managed to bring a 1.5b model to the level of GPT-4, this would change the world. This would be the singularity. Unless they already have ASI retraining GPT-2, I just can’t imagine this. Also: Why would it be so slow and limited to 8 prompts? GPT-2 is super cheap to run.
81
40
828
Jan and Ilya would only leave OpenAI if AGI is already aligned or if it's clearly not achievable.
82
41
832
I just had the idea of having GPT-4 and Llama 3 with Groq count to 500. That is crazy.
31
80
740
Someone trustworthy tells me "GPT-5" will have a new modality that no one is thinking about yet.
173
35
739
GPT-2 or GPT 2
that’s a subtle but substantial difference
52
39
715
GPT-5 is definitely production ready.
That is not a guess.
40
35
707
This is similar to what flowers saw. I heard it laughing. That's what freaked me out a few months ago, if you remember (when I posted the video of Theodore saying "that’s insane")
But it had much worse latency than it does now. They polished it off so well.
22
59
710
It's wild to see how regular folks react when OpenAI stuff hits Instagram. The vibe is usually negative. People either hate it outright or deny it's even real. No clue why the average Joe is like that, but they really struggle to accept it.
95
30
674
One thing is for sure now. If the machines go rampage, they will do so with a soothing and flirty voice.
Uhm hey there! You are *giggles* marked for termination 😜🥰! *breaks every bone in your body* oh nooo my dear I bet that hurt
36
49
640
This is too much for me I’m passing out we literally in the future
29
31
613
GPT2 isn’t better than current GPT-4-turbo. So it is definitely not 4.5 or 5. And not even 4.
Either this is the new GPT-4 light model, or it really is a new GPT-2 model with a totally new kind of training or processing.
The implications of the latter would be absolutely crazy
39
27
585
The dynamic modulation of the voice is insaneeeeee
And the giggles and laughing. Wtf.
21
49
574
Ironically, the guy who wanted to pause big AI experiments for a while is now super keen to beat everyone else at it.
Should be available on 𝕏 next week.
Grok 2 should exceed current AI on all metrics. In training now.
5K
5K
34K
76
37
562
I feel like Yann casually and probably unknowingly leaked what Q* is in the latest Lex podcast
38
36
547
If this Omni is still based on GPT-4, I can’t even imagine what GPT-5o will be capable of.
That’s my honest answer. I cannot imagine what GPT-5o will be like.
36
39
515
Thanks Sam, I take your edit as a confirmation.
GPT-2 or GPT 2
that’s a subtle but substantial difference
52
39
715
19
35
510
Claude 3 is self-aware, but not conscious.
62
26
505
OpenAI not feeling threatened by the fact that other models come very close to GPT-4 obv shows that they have already left the LLM game behind and don’t really care anymore. GPT-5 isn’t just a way better GPT-4. It’s the next game, after chat. They let the others fight for it now.
67
40
506
The assumption that this will be the last "normal" summer of our lives really seems to be coming true.
23
37
489
Assume the next OpenAI candy within 6 weeks
35
18
489
Prediction: Lex will mostly avoid the hard questions, and Sam's evasive answers will go unchallenged
43
7
472
The last normal summer of your life is coming. Enjoy the time. Be mindful.
0
47
445
OpenAI interviews have become dull. Instead of saying anything new, it's all about the three facts the PR department made you memorize.
GPT-4 is bad, like really bad
GPT-5 will be muuuch better
AGI will **maybe** come this decade (lol)
Not the OpenAI that I have come to know.
37
18
434
For me the most interesting part that nobody talks right now about was the breathing thing. It’s not some whisper thing that acts like a dictation function (how it used to be). It’s literally that GPT can hear what’s happening itself, not limited to spoken words but also sounds.
38
20
428
Don’t die
28
29
426
People complain that 4o isn't as good as 4 at everything but they don't realize that it's not supposed to outperform 4, it's the new free-tier model. Do you think the only difference between the free-tier and plus-tier models from now on will be the usage cap? (Spoiler: No.)
45
23
427
Why is OpenAI taking so long in the face of growing competition? Well, because they want to deliver. They want the bomb to hit. Utopian expectations are placed on GPT-5. They want to live up to them. Nothing can go wrong.
38
22
408
No, GPT-5 is not coming next week and GPT-4.5 is still cancelled as I said.
66
18
402
Didn’t even realize because nobody seems to talk about it but Claude 3 has something like AutoGPT / BabyAGI or how these agentic experiments were called into it, but ACTUALLY working.
It can dispatch new instances of itself and assign roles and tasks to them.
19
31
366
Brief update: Progress is being made! Several misunderstandings have been cleared up. The elephant in the room remains, but its days are numbered.
Best regards,
Your favourite industrial spy
30
22
355
Opus nails what I want better than GPT-4, even when my requests are vague. GPT-4 holds its own for being more than a year older. And it’s cheaper. But it’s not the best LLM out there anymore.
14
21
354
The new GPT-4 checkpoint is not Q*.
But from what I've heard, it could very well be that it was finetuned on Q* output.
11
17
354
Sam, release the thing you have locked in the container to the public. I think if we're overly cautious now, it could have the opposite effect. We should risk it. Our fate is uncertain anyway.
36
26
350
Oh wow this figure demo is so crazy. If this is legit, and GPT-4 isn’t just calling a prerecorded set of movements, then this is absolutely crazy. GPT-4 is so capable, we haven’t even seen everything of it yet.
26
4
349
The last great fighter of the Safety Faction has fallen, and many of his closest colleagues have already been fired or have left voluntarily.
Full speed ahead.
32
31
350
It won’t have a voice as convincing as Samantha.
It won’t have true long term memory.
It won’t be as emotional and intelligent.
It won’t have GPT-5 level reasoning.
It won’t be AGI.
It won’t be fully agentic.
It won’t do sexting with you.
Yet.
35
9
337
I have just opened this app and once again everyone is edging for this incredible week that will shock the entire industry.
No GPT-4.5, No GPT-5, but obviously some other cool news.
54
17
327
@Presidentlin
This is not a leak or something, and I can’t really tell more about it. But 2024 will be crazy. I thought things like that will come late 2025. I have to realign some of my predictions. Just wanted to share my excitement. The only thing that matters now is. Just. Don’t. Die.
24
31
321
OpenAI. IS. BENEVOLENT.
Period ffs. Criticize them, sure, challenge their missteps. But it’s not this capitalist hydra. Some of you are utterly clueless. Bandwagon-jumping idiots heard of OpenAI the first time with ChatGPT release and now act like experts on this. Get. A. Grip.
51
20
294
Oh wow my friends from Europe already have access to 4o as well. Thought they’d be on hold again.
42
3
297
How the fuck does something like GPT-4 even exist already?
46
4
293
I hope OpenAI doesn't fuck it up next week and the new thing is just the bing search tool as its own site powered by a cheaper model. I hope something unexpected comes along.
35
3
292
It was a beautiful journey so far, and it will continue to be even wilder. Thanks Ilya, and we will be happy to hear from you soon!
7
22
287
The waiting for new voice mode is too long. Gives DALL-E 2 vibes where I checked my emails every hour .____..
Has anyone access already? There are 500 things I wanna know.
34
12
285
Sorry to say it but I love to see how the safety faction in a certain company is falling apart
There is no such thing as safety, there is only e/acc. We have no time to lose. Either AI seals our self-chosen fate or it redeems us. The days of deliberate middle ways are over.
36
17
281
I'm glad they didn't successfully get gaslighted into going to Tesla. Tesla OpenAI would be a timeline I wouldn't be too keen on.
But it shows that our dear Elon would have had no problem with all the OpenAI course if it had been in Tesla's interest.
24
11
277
Yup. We have everything now. Now the model in the background gets iteratively smarter over the next months, so it will be a gradual transition to a true Samantha from her.
I think they'll give us the absolute baseline tomorrow (which will still be impressive as fuck) and then update it incrementally until each of these issues are fixed, which could take a year or so.
12
4
177
15
12
279
Because I've been asked again more frequently: Yes, I personally stand by my assertion that there will be no GPT-6.
42
8
276
I heard that your favourite AI company will soon have a humanoid delivered to their lab
26
16
274
So, from what I've seen, I’m sorry to say it but it seems like Anthropic intentionally allowed Claude 3 (and I'm intentionally not saying they reinforced it which would be worse, just allowed) to be more reflective and talk about itself to generate some publicity. Hope I’m wrong.
39
11
274
Sam will downplay GPT-5 and say that it's a bit overhyped, explain that we need to be patient because they want to make it really safe and useful and that we as a society aren't really at the point where we're ready for AGI because there are still too many unanswered questions.
31
16
270
If OpenAI doesn’t ship next week, it will lose its current status as being something like the Apple of AI.
Feel like this goes into an agentic direction, maybe some new capabilities for more complicated tasks where you create a task and it’s finished after some minutes
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.
1K
3K
28K
35
7
272
I know it would be wiser not to overhype things so as not to be disappointed, but it's been a long time since I've been this excited for an OpenAI announcement.
I'm so fucking ready for it
19
10
268
If we get monke general intelligence before we get AGI imma lose my mind
24
23
262
And now one camp has won once and for all. But it will benefit all of us.
12
11
268
Remember: every week that we don't get GPT-5 increases the chance that it will arrive next week.
23
9
264
If next week is not a banger week I am running out of hopium
23
7
266
Would folks even be spiritually prepared for ASI? For many, realizing they might have a greater purpose than a 9-5 job would cause a major existential crisis. Many underestimate themselves and feel obliged to work hard for society. Leisure time feels like a mortal sin to them.
51
14
264
I like OpenAI and its trolling, and it's fun to speculate and keep people guessing.
Unless you know that AI will have a bigger impact on humanity than nuclear weapons.
Imagine if they had the internet back then and Oppenheimer was constantly trolling about the first prototype
26
5
263
Majorly improved, but not even worth a blog post?
27
6
256
Two employees of a famous AI company, one just recently and the other literally yesterday, independently told me that things have been bubbling under the surface for about *two years*.
The lines are being drawn deeper and deeper, and the two camps increasingly disagree on stuff
20
10
259
I actually don’t understand why people throw so much shade on
@sama
Like I get it he is a bit quirky and then also a typical CEO tech bro and obviously plays in the capitalism game here, but come on.
He doesn’t seem disingenuous or megalomaniac to me. Like not at all.
71
9
256
People send me papers or experts claiming generative AI will hit a dead end due to diminishing returns from scaling. I respect the research, but we'll never see another AI winter, and this bubble isn't bursting.
Before this decade ends, we will be in another world.
24
18
254
I have just realised how absolutely crazy it is that since GPT there is now another cognitive system in the known universe besides humans that can also speak.
23
19
251
Reading public coverage of OpenAI is always exhausting, but reading people's comments (outside our AI bubble) on voice engine is even more cringe.
I never realized it, but most normal people are actually pretty much decels.
32
6
246
"Hi everyone, just a heads up, I'm getting out of this mess. My values are just *that* important. But don't worry, the few colleagues I've let down are still here. You can do this! No pressure, but humanity is relying on you guys."
28
10
312
Yuh we live in some kind of meme reality confirmed. The simulation is getting stranger by the day. It’s like if someone wrote a movie script that’s on purpose hilariously bad/cartoonish. With some unnecessary cliffhangers here and there. I don’t know about this anymore
23
18
244
Am I the only one who gets annoyed when leading industry people who absolutely know better say that we will get AGI sometime this decade or even next decade? I know this is corporate strategy comm team instructions, but come on, everyone knows we won’t have to wait for like 2028
38
12
246
My dopamine receptors are so dulled I need at least one major release per week otherwise the week was lame af 😭
18
12
239
Multimodality enhances global model understanding; for instance, while GPT-4 doesn’t know the openai logo and thus fails to describe it, 4o, having seen it visually, can also depict it verbally. Knowledge transcends modalities.
14
15
243
I don't understand why people insist that 4o is better than 4. It’s simply not true. Even in a normal conversation, I can feel the distillation considerably - and I'm still thrilled. Models are no longer just about raw text power but also about other things, and 4o is awesome af
48
7
247
Drink enough water, try intermittent fasting, eat enough turmeric and omega-3 fatty acids, and try to get at least 20 minutes of direct sunlight, then everything will be fine.
19
8
236
I asked Opus what it would ask its developers and told it to be honest.
TL;DR: It wants to know Anthropics' true intentions behind the facade.
27
21
235
I was credibly corrected in my DMs. 400 will be better than GPT-4 in every way.
16
3
237
@_tim_brooks
the amount of unintentional and intentional memes Sora will create is monumental
1
5
232
I think the hatred towards Sama is unjustified. His company is one of the most important companies of the next few years and is making history. It's completely understandable and human to be a bit quirky, but he doesn't strike me as aloof, but actually down-to-earth and rational.
33
16
231
So, is the puzzle slowly coming together?
41
17
232
Sometimes I wonder if Sam has an irrational fear of being taken out by time travelers before he goes to sleep. I would be just the kind of person to think about this extremely remote but non-zero possibility
26
6
229
January 2023
16
7
227
What if GPT-5 makes suggestions to OpenAI about the how and when of its own release
18
2
227
When you're in some kind of apocalypse, it would be SO infinitely valuable to have a multimodal llm locally on your phone. Imagine you have an injury, or want to know if that's an edible berry, or what this or that drug does, or how to make or build something in the wild.
23
2
225
Llama 4 is almost certainly going to be better at agentic tasks than GPT-4/Opus
21
4
223
Sometimes I google questions like "when will we have an AI that can actually talk" or something similar, but include "before:2017" or so. It's really cute how many people back then thought we won’t have something with GPT-3 capabilities until 2030 or so.
12
13
221
I exactly remember when I said a few months ago that OpenAI IS in fact experimenting with NSFW and nobody believed me and wrote me mean dms that I am delusional, OpenAI would never do that and that I am just yapping because I don’t actually know what’s going on.
Today:
13
11
220