@sama When they talk about “slowing things down,” what they usually mean is slowing down access for end users like us.
I don’t believe model development will slow down, or that they’ll stop benefiting from this technology.
They’ll just keep the best of it to themselves.
When they talk about “slowing things down,” what they usually mean is slowing down access for end users like us.
I don’t believe model development will slow down, or that they’ll stop benefiting from this technology.
They’ll just keep the best of it to themselves.
I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks.
Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon.
Sometimes App Store review takes 12 hours, sometimes 14+ days. I understand busy periods. But what I don’t get is this: if an app gets rejected in 12 hours, I fix the issue immediately and resubmit, why can the next review take over 14 days? How does that make sense?
I think we’re entering a strange phase of the internet where the people who experiment the most will have a bigger advantage than the people who “know the most.”
AI changes too fast for expertise to stay static.
@benfurkankilic https://t.co/P7RB1ttgzd ve Fireflies’ı bu amaçla bir süre boyunca test etmiştim. Açıkçası kişisel toplantı yönetimi gelişimi açısından https://t.co/P7RB1ttgzd daha başarılı gelmişti bana. Türkçe desteği ve detaylı raporlaması da cabası. Entegrasyon da ararsanız FF daha iyi gibi.
@chiefkittenme Treat them like your second child. You know how everyone is super careful with the first one, then by the second they’re way more relaxed and slightly more “eh, they’ll be fine” about everything.
Or just have a second kid ASAP 😂
GPT-6 Astra is really good. But right now, it’s just too expensive.
So far, I haven’t been able to complete a single prompt in one session. About 10 minutes after I submit a prompt, I hit my 5 hours usage limit before the task is even finished.
How’s your experience been?
Same thing with AI.
Yes, those impressive demos are absolutely possible.
But the real question is: with what prompt?
Access to the model is becoming a commodity. Knowing how to get the right result from it still isn’t.
With every new AI model, we see posts like: “Look what it built from a single prompt.”
But that’s a bit like building Tetris in Excel.
Everyone has access to Excel. That doesn’t mean everyone knows how to make their way into building Tetris with it.
I’d already automated most of my work anyway.
All I needed to do was check the final output and approve a few things. And for that, my phone was more than enough.
Being able to keep things moving without carrying a laptop around felt ridiculously good. 😄
I’m loving the remote-control features in AI apps lately, especially being able to use Codex Desktop and Claude CoWork from my phone.
I went on a one-week vacation recently and really didn’t want to bring a laptop, so I just left my computer on at home.