News for hackers. Stuff that matters.
Re:Google dusts off the failed Google+ playbook to fight Cha... by matheusmoreira on Friday March 10, 2023 @10:28AM (#35095556)
We really need the power to run these things on our own computers with 0 restrictions. Not these bastardized corporation friendly versions.
Re:Google dusts off the failed Google+ playbook to fight Cha... by LadyCailin on Friday March 10, 2023 @11:09AM (#35096240)
It requires 300GB of RAM apparently, not sure how you propose to make that work.
Re:Google dusts off the failed Google+ playbook to fight Cha... by marcyb5st on Friday March 10, 2023 @11:37AM (#35096723)
And specialized hardware. Not sure how long an inference on CPU only will take, but I suspect it will crush usability.
Re:Google dusts off the failed Google+ playbook to fight Cha... by matheusmoreira on Friday March 10, 2023 @11:39AM (#35096752)
Well, my laptop has 64 GB. It can't be that difficult.
Re:Google dusts off the failed Google+ playbook to fight Cha... by pixl97 on Friday March 10, 2023 @04:45PM (#35101541)
This is one of those comments you can't tell if its sarcasm or not...
Re:Google dusts off the failed Google+ playbook to fight Cha... by ojosilva on Sunday March 12, 2023 @06:55AM (#35118651)
We said the same thing, in one way or another, about indexing and searching the web -- privacy/anonymity, spam/seo, algorithms, bias, corporations.... So today one can build their own crawler and search engine with great opensource software, yet a vast majority is querying a web index through third parties as the computing power required (well, mostly storage and networking) remains brutal, and then all those nitty-picketty details about crawling one needs to painstakingly figure out.

Even if by Moore's law we get that great power in our own computers, enough to run a today's chatGPT clone, there will always be a commercial version that is 100x more awesome out there built by actors with immense compute power backed by rivers of corporate money.