HackerTrans
TopNewTrendsCommentsPastAskShowJobs

julkali

433 karmajoined 8 lat temu

Submissions

[untitled]

1 points·by julkali·6 dni temu·0 comments

Google Translate apparently vulnerable to prompt injection

lesswrong.com
59 points·by julkali·5 miesięcy temu·3 comments

Reticulum Network

reticulum.network
2 points·by julkali·6 miesięcy temu·0 comments

Recessions have become ultra-rare. That is storing up trouble

economist.com
3 points·by julkali·7 miesięcy temu·2 comments

Programming built the tower of Babel, AI destroyed it

video.ethz.ch
3 points·by julkali·8 miesięcy temu·2 comments

comments

julkali
·5 dni temu·discuss
Please support your claim about (networking) performance of BSD-based systems and Linux with some source(s). It surprised me. Thank you.
julkali
·23 dni temu·discuss
nimby
julkali
·w zeszłym miesiącu·discuss
no
julkali
·4 miesiące temu·discuss
i felt that.
julkali
·4 miesiące temu·discuss
because you start internalizing costs
julkali
·6 miesięcy temu·discuss
Source?
julkali
·6 miesięcy temu·discuss
I also tried setting my phone to grayscale, but it takes me significantly longer to do useful things, as some UIs are harder to distinguish without color. Have you found a workaround for this?
julkali
·6 miesięcy temu·discuss
Yes, but only 15-20 years later in the extent expected for 2000. The technology wasn't there yet, just like with LLMs.
julkali
·7 miesięcy temu·discuss
https://archive.ph/i0sFU
julkali
·10 miesięcy temu·discuss
The issue is that the examples you listed mostly rely on very specific machine learning tools (which are very much relevant and good use of this tech), while the term "AI" in layman terms is usually synonymous for LLMs.

Mentioning the mid-1990s' internet boom is somewhat ironic imo, given what happened next. The question is whether "business models mature" with or without a market crash, given that the vast majority of ML money is provided for LLM efforts.