Hacker Newsnew | past | comments | ask | show | jobs | submit | egillie's commentslogin

I'm seeing it outside of techies. My dad told me "AI Google said that..."

most of what we know today we knew years ago, too


Is this because GPT-5 hallucinates less in general?


Explains why they’d announce on a Saturday


Where did you find the reporting about the non consensus ai view? That’s the most interesting take I’ve heard on this


And scale doesn’t even have the best data among these vendors so I also don’t get this argument


What are some other options ?


Good one Zuck.


Inodata


All of my passwords are encoded in pi…somewhere


Well, probably! But it hasn't actually been proven yet that pi is a "normal number" [0], though most mathematicians think it must be.

[0] https://en.wikipedia.org/wiki/Normal_number#


That's a stronger claim anyhow. We'd merely need Pi to be a rich number, ... which hasn't been proven either.

https://en.wikipedia.org/wiki/Disjunctive_sequence#Rich_numb...


It's not necessary that pi be normal in order for it to contain every possible sequence. Sufficient, but not necessary.


If the OP is anything like me, then they have no trust in number theory researchers and deliberately choose passwords like "141592" so they can be sure


I enjoy thinking about this. Not only is a sequence of digits encoding the complete works of Shakespeare in there somewhere, it's in there again. And again, an infinite number of times.


In my encoding I represent the works of Shakespeare by the digit 0, anything else by the digit 1 followed by its ASCII representation.


This makes me wonder if you could train an llm without any references to Wiles’ work and see if it can compete Fermat’s last theorem


I’ve heard it’s all the crypto influencers jumping into the AI hype


Could even argue that the price should go up, since the amount of with one GPU can do and its potential ROI just increased


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: