People get an answer, and not the answer. It seems that people are fine with this. Even this article's example answer is false. The author didn't care, and considered it good.
I believe that Google has figured out (correctly, IMO) that accuracy doesn't matter to most people 99% of the time, and people will likely do a deeper search for the 1% of the time they do want it.
If people really wanted the truth and facts, we would not have misinformation spread this widely via social media and other places.
People prefer to be lazy and sit and do nothing, than to move around to achieve a goal.
Is that a good thing? The reality is most humans are becoming more and more intellectually lazy - as a result their cognitive function are in decline. Therefore if something looks right at face value / supports an internal bias - they take it and run with it.
It is definitely because it's a smaller model. TPUv7 has ~10% lower flops at FP8 and 33% lower memory bandwidth than Nvidia Blackwell cards. Add CUDA to the comparison and they'll probably be even worse at real world utilization. Grok is already running on Blackwell cards and although there's little info on GPT5, I doubt they are behind.
