>>106271458
Delusional Futurist types thought that.
The sort of fucktards that watched the news and did normie TV science for pea-brains.
We're doing things right now we thought were impossible just a decade ago.
Get a fucking clue and stop being a dumb child.
>>106271542
>you mean this AI with no understanding of human emotion that just regurgitates the most likely outcomes is shit at writing????
Try do the same with thinking mode on 100% of the time and you'll see a difference. Good luck paying for it.
Expensive != impossible.
GPT5s demo had much MUCH more resources given to its test subject-- sorry testers and the instant it was released it was in a horribly gimped form.
They're not going to give people full unlimited access to stupidly expensive models. They can say "our latest model is our cheapest yet!" all they want - they're still not making shit on any of these things, running entirely off of investment.
But to deny that these things will improve is patently wrong, and downright retarded.
They've only BEEN improving.
They hit a ceiling of COMPUTE, not model improvements.
That forced them to rethink how the models worked and it only took that Deepseek company in China to shake the entire industry with a total redo of how Transformers were processed and every company followed in their footsteps, leading to huge improvements in all of them.
