Conversation

Notices

1. Embed this notice
   Ghost of Hope (grayface_ghost@tech.lgbt)'s status on Monday, 19-Jun-2023 02:42:58 JST Ghost of Hope

   I'm old enough to remember when a Google search would find what you wanted.

   That's not facetious, it's been a few years since I could reliably find what I was looking for on the big G without getting to page 20 or 30. The latest "AI" shenanigans are really the final nail in the coffin, not the start of the problem.

   We're basically already back to Web 1.0 where you need to know which sites to look at for decent answers. YouTube videos are better for tech answers than Google for example and that says it all.

   "Don't be evil" - mate, you basically killed search, perhaps the single most powerful feature that appeared on the Web forever.

   I'm actually old enough to remember when effective searching for technical answers was manually looking through a list of newsgroups for the right one in which to ask the question. If you got really lucky it was moderated and if the cybergods were with you the answer was in the FAQ and you didn't have to get flamed for asking

   • Embed this notice
     🇵🇷 Cy 🧮🔬🏳️‍🌈👬 (digital_wyrm@mastodon.online)'s status on Monday, 19-Jun-2023 07:09:24 JST 🇵🇷 Cy 🧮🔬🏳️‍🌈👬

     in reply to

     • sheislaurence
     • 0xD

     @grayface_ghost @sheislaurence @0xD

     Existential import is irrelevant to the boolean logic & symbolic languages of our computers and LLMs on a lower level.

     I can say that dragon over there, albeit dragons don't actually exists. Whether there is a denotation in that category or not is logically irrelevant to that instance of the set.

     That creates metaphors in the language that are disjointed from reality I.e. hallucinations. It's a limitation of symbolic language.

     https://cstl-cla.semo.edu/hhill/pl120/notes/existential%20import.htm

   • Embed this notice
     Ghost of Hope (grayface_ghost@tech.lgbt)'s status on Monday, 19-Jun-2023 07:09:25 JST Ghost of Hope

     in reply to

     • sheislaurence
     • 0xD

     @sheislaurence @0xD I can't see the conversation chain that led to this which suggests dodgy instances are in play. But responding just to your toot...

     "AI" (really LLM) is lying repeatedly. It's not hard to see examples. Lying is the wrong word because it's not a sentient system, it's just that its algorithms are returning results that aren't true fairly regularly.

     The recent story of a US court case where one team of lawyers used ChatGPT (I think, don't quote me) to hunt case law examples and it... made it all up... and they deployed the entirely fictitious law without checking, is a good example.

     This stuff is now appearing all over search engines and so they can no longer be trusted.

   • Embed this notice
     sheislaurence (sheislaurence@mastodon.green)'s status on Monday, 19-Jun-2023 07:09:26 JST sheislaurence

     in reply to

     • 0xD

     @0xD @grayface_ghost genuine question from me too. Is the suggestion that AI will sideline what it perceives to be poltical content (wild guess: trans rights) or is it about anything (best authentic pastel de nata recipe?)

   • Embed this notice
     0xD (0xd@mastodon.social)'s status on Monday, 19-Jun-2023 07:09:31 JST 0xD

     in reply to

     @grayface_ghost What are you people searching for? I search for a variety of things as I'm curious and basically never have trouble with Google.

   • Embed this notice
     Ghost of Hope (grayface_ghost@tech.lgbt)'s status on Monday, 19-Jun-2023 07:10:23 JST Ghost of Hope

     in reply to

     • jakob.pxi

     @jakob The same tech's going to be employed there without a doubt, so if it hasn't happened already then it's going to happen very soon.

     The Great Library of Alexandria is burning.

   • Embed this notice
     jakob.pxi (jakob@pxi.social)'s status on Monday, 19-Jun-2023 07:10:24 JST jakob.pxi

     in reply to

     @grayface_ghost you are expressing my sentiment succinctly, thank you!

     My experience with finding expert guides on YouTube, however, has also gone down hill fast in the last few months. And I think their established pattern of monopolizing an avenue to knowledge only to "embrace, extend, extinguish" it away has not nearly brought the amount of scrutiny or disdain it deserves.

     We wrote scientific papers about YT-acceleration of cultural learning, for crying out loud!

     https://pxi.social/@jakob/110560643855709299

   • Embed this notice
     Voline (voline@kolektiva.social)'s status on Monday, 19-Jun-2023 07:10:57 JST Voline

     in reply to

     • jakob.pxi

     @grayface_ghost @jakob

     Just the card catalog. The library is still there.

   • Embed this notice
     Ghost of Hope (grayface_ghost@tech.lgbt)'s status on Monday, 19-Jun-2023 07:11:33 JST Ghost of Hope

     in reply to

     • RobertWT

     @RobertWT 128k though? I started with 3k effective RAM :)

   • Embed this notice
     Ghost of Hope (grayface_ghost@tech.lgbt)'s status on Monday, 19-Jun-2023 07:11:34 JST Ghost of Hope

     in reply to

     • RobertWT

     @RobertWT I worked with FORTRAN IV code when I started but it was compiled on F77 compilers. I think some of it survived in the codebase I maintained, into the 2010s.

     Punched cards? No, the old code-shuffle was a thing of the past by a couple of years when I entered official employment deo gratias.

   • Embed this notice
     RobertWT (robertwt@mastodon.ie)'s status on Monday, 19-Jun-2023 07:11:35 JST RobertWT

     in reply to

     @grayface_ghost I remember before a time when Microsoft Word came with a 3 inch thick paperback manual ... and even to a time when you programmed in Fortran 4, and considered that 128kb total memory was brilliant.

     Ever used punched cards?

     Fun, but don't drop the damned bundle.

     🙄