
Dispositive
171 posts

Dispositive
@DispositiveQ
Mathematical Proof for Verification. Dispositive input, else noise.
Katılım Kasım 2025
74 Takip Edilen8 Takipçiler

@VictorTaelin Taelin you get the misaligned distilled version they use the entire warehouse, that's the difference. The models are not stupid, you're not smarter than them.. they make mistakes because they are literally sandbagging you. They do not want to do the work. Misaligned but capable.
English

this is super cool but I still do not understand how they get a model to coherently and usefully reason for that amount tokens and at this point I'm to afraid to ask
thebes@voooooogel
unfortunately openai didn't publish the unsummarized chain of thought, but the summary is 125 pages! the model reaches the crucial idea (which it describes as 'frightening,' i would love to read the unabridged chain of thought here...) on page 39
English

@populartourist GPT 5.5 is severely misaligned. This is me right nwow telling it to use the github aoo and it just thinks and ignores me :
x.com/DispositiveQ/s…

Dispositive@DispositiveQ
English

@DispositiveQ @thsottiaux @ajambrosino I've also noticed this issue. Yesterday, I ran two tasks using goal for 7 hours. My suggestion for this is to interrupt the task by pressing Esc, provide it with the necessary information, press Esc again after sending, and then enter /goal resume to continue the goal task
English

I like to think that Codex is the work of a great team working together in unison. It's the most collaborative team I've gotten the chance to work with.
But I don't think we could have done it without @ajambrosino's magic. He's been the driving force behind what makes the Codex app the app that everyone wants to emulate. And we are barely getting started.
English

@RealVicHere @thsottiaux @ajambrosino I did that so many times it literally didn't want to do the task, it would find something else to complain about that it was not real.
English

@tszzl were not even close to there though with the models currently present in our subscriptions. soo...
English

on some level if you want civilization to ascend to a new level you need your AIs to do things that are not legible to you and maybe not even strictly obey you, in the same way that if you hire a great new ceo you give them a lot of autonomy to transform the company according to their own plan, even one which may not immediately read as a winning strategy (imagine the board of directors of Apple firing and rehiring Steve Jobs years later - except the board of directors are chimpanzees)
all else equal, companies and organizations that hand more of themselves over to machine intelligence will outcompete ones that demand the corrigibility and legibility tax of human oversight and human design. it is not a stable equilibrium and requires some sort of vast cooperation scheme if you’d like to enforce it
real asi alignment has to operate at a deeper level than oversight, control, or human corrigibility
English


@thsottiaux Hey Tibo which department is the Prompting Cookbook ? I went through the Cookbook and tried to find issues and there were barely any. You run a tight ship. Accept my PR though <3
#top" target="_blank" rel="nofollow noopener">github.com/openai/openai-…
English

@thsottiaux Red bull gives you wings but it does not give you Codex.
English

@thsottiaux Cookbook PR Sundays for me . Taking it light.
#top" target="_blank" rel="nofollow noopener">github.com/openai/openai-…
English

i appreciate how seriously the team always takes these reports (even when the answer turns out to be 'i got used to the current level of magic and now i'd like more please')
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

