@Smilogeek "metadata about the app" like what, the app name? it's mostly unnecessary info outside of what is seen in the screenshot already. the entire result is virtually the same, it pastes a screenshot in Codex, plus the name of the app
@scaling01 Yeah he's waiting until most of the users are near the natural weekly reset (it resets in one day for me) and then it hits reset few hours before that in order to claim the PR benefits of being a god. Such a fraud
@mario_raion@AggrendWoW Yeah and these don't matter for raytracing or global illumination as we know from the other modern games. We know adding path tracing to Quake 2 reduced the fps from 1000+ to 30 fps at the time.
@mario_raion@AggrendWoW What do you mean "crazy"? Haven't you seen the rendering panel from Blizzcon? It's the retail engine, with volumetric shadow that casts shadows, global illumination, raytracing and all new liquid simulation with waves and reflections. Retail doesn't have these yet
@White1637402 I think you can understand that both can happen at the same time. The entire "pacing" thing only works if China also agrees. No China agreeing, there will be no pacing, *obviously*
@mattshumer_ In 2020 people were dying literally (highest mortality rates outside of war times) many more being sick, hospitals full. The AI risk is not immediate and is not visible, it's the risk that bad actors (Russia, China, Trupm) could use nukes to kill entire nations
@DrKr00@testingcatalog 99.99% privacy breaking (illegal in most of the countries, even) while being 0.01% useful for a small fraction of the user base in select few scenarios.
@testingcatalog Excuse me but every other Apple Watch was "able to capture audio from the world around us", they now just made it spy on us 24/7 while selling it as some sort of useful feature. Give us back the battery wasted by having the microphone recording 24/7...
@nisxant@DanDr1s They won't have unique names per each generation, they specifically mentioned that the model name will show the parameter size: Luna (smallest) < Terra < Sol < Astra (biggest) -- it's exactly like Haiku < Sonnet < Opus < Fable, you need to get used to this
@VDN_00001@benedictk__ What would "the answer" be? They just released the new Fable two days ago, their best, most advanced and most expensive model! There's nothing else better than that they are releasing now. Maybe Opus 5.1 if they hurry it but we don't expect it to be better than Fable 5.1, right?
@DanDr1s They've said that they've fixed the frontend with 5.6, then once again when teasing Astra. It's pretty obvious by now that they just don't know how to do this, Claude is so far ahead on this domain
@DanDr1s I think what happened in the past was that models were trying to create things from memory and now with 5.6 and 6 they're literally become better at copying stuff from a reference image. 6 is exceptional at copying, but not that good (below Fable 5.0) at creating from scratch
@ChrisGPT Because he's saying things like this every 2 months since 2-3 years already, every time the next model is going to be transformative, a huge increase and so on. It became the norm, it's why he's being ignored.
@scaling01@ege02 MOST of what he posts is speculation, makes him look bad. He was wrong on most of what he said in most of the big AI model releases for the past 6 months. Even more so he acts so dumb. Insults Dan, saying he's an AI slop account, yet Dan correctly leaked several releases
@reach_vb Indeed a small but nice quality of life feature! But you know what would be actually huge? Us getting access to the already "released" but not-yet-released GPT-6. Imagine doing a launch event without launching anything!!!! All your users are fuming, such an embarrassing moment