Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
post hole digger
Mar 21, 2011
bing chat: no one would even care if i died. might drive home from the bar tonight..................,,,

Adbot
ADBOT LOVES YOU

echinopsis
Apr 13, 2004

bssoil
Mar 21, 2004

Malloc Voidstar posted:

bingGPT going well





hosed up, did it ever tell you when Avatar was playing?

Beeftweeter
Jun 28, 2005

a medium-format picture of beeftweeter staring silently at the camera, a quizzical expression on his face

bssoil posted:

hosed up, did it ever tell you when Avatar was playing?

avatar 2 has not yet been released

echinopsis
Apr 13, 2004

Cybernetic Vermin
Apr 18, 2005

in the interest of saving the cyberpunk thread from this, without just shutting down conversation, i will state the entirely unironic and honest belief: if natural language processing is a thing that can possibly be defined distinct from general artificial intelligence, that has probably now happened.

which is probably not a very interesting statement to most people.

endlessmonotony
Nov 4, 2009

by Fritz the Horse
Argh if it's not breaking down the meaning behind your words and figuring out how to control the underlying systems, it's not natural language processing it's a complicated button! If you still have to have an understanding of the underlying systems it's just an interface like any other!

That being said, by being bolted onto a specific system cleverly enough it could do natural language processing for specific use cases. Y'know, the way it can't now, what with being very confidently entirely wrong a lot, too much to rely on.

Unless it can un-gently caress its own mistakes without human prompting, double-check when it's unsure, communicate its lack of confidence... it doesn't understand a goddamn thing. It's not thinking.

Though to get there I think it might need frustration, anxiety, and self-doubt. And if we're programming a thinking, feeling computer and giving it all these negative emotions, that smells like the Torment Nexus to me.

Cybernetic Vermin
Apr 18, 2005

that it is a perfectly valid position that nlp short of gai is impossible. it used to be a very common, possibly majority, opinion, but it has eroded a bit of late. i honestly can't argue about it, while i believe we've now found a meaningful tier short of gai, it is not like it is objectively established.

endlessmonotony
Nov 4, 2009

by Fritz the Horse
NLP requires a model of the systems it's interfacing with. AGI requires a model of itself it's capable of evaluating and changing.

endlessmonotony
Nov 4, 2009

by Fritz the Horse
Presumably there's a fuckload of other problems to solve too, but the system consistency evaluation is a requirement we're not capable of doing. We do have tests for it, and the way machine learning works, as soon as its training data has people discussing the test, the test is useless.

Moo Cowabunga
Jun 15, 2009

[Office Worker.




machines need operators.

I learnt that and many other lessons on the farm.

Agile Vector
May 21, 2007

scrum bored



College Slice
the content farm :dadjoke:

SpaceAceJase
Nov 8, 2008

and you
have proved
to be...

a real shitty poster,
and a real james
A useful AI tool would be to listen / read random notes and create some structure from them.

Like some random string of brainstorming or a way to categorise notes from a meeting. Where are the tools that can do that and don't look like scams?

FalseNegative
Jul 24, 2007

2>/dev/null

SpaceAceJase posted:

A useful AI tool would be to listen / read random notes and create some structure from them.

Like some random string of brainstorming or a way to categorise notes from a meeting. Where are the tools that can do that and don't look like scams?

Microsoft claims Bing will suck in the voice transcripts and do this with Teams Premium:

Microsoft posted:

ChatGPT acts like an AI secretary for Teams meetings, similar to Otter.ai, Airgram, or Supernormal. The AI automatically transcribes and summarizes the conversation, pulling out key comments and offering a tentative schedule of future events and action items based on the discussion. The Microsoft Teams recap page will include a recap of the meeting, with speakers marked in the transcript and the whole thing divided into chapters by the AI and applicable to PowerPoint Live meeting recordings as well as the standard meeting format.

https://voicebot.ai/2023/02/02/new-microsoft-teams-premium-uses-chatgpt-to-take-meeting-notes/

distortion park
Apr 25, 2011


slack does autotranscripts of some videos now and it's very bad

Agile Vector
May 21, 2007

scrum bored



College Slice

distortion park posted:

slack does autotranscripts of some videos now and it's very bad

Cybernetic Vermin
Apr 18, 2005

distortion park posted:

slack does autotranscripts of some videos now and it's very bad

i expect this will be real bad too. summarization is increasingly not terrible, but that's running on like coherent edited texts, doubt it'll understand or prioritize things sanely in a poorly transcribed freeform conversation.

has a lot of room to be hilarious though.

Beeftweeter
Jun 28, 2005

a medium-format picture of beeftweeter staring silently at the camera, a quizzical expression on his face
if it's based upon their bing/sydney model then i'd expect it to also be sociopathic, though that may be more difficult to discern in a business setting

Cybernetic Vermin
Apr 18, 2005

stuff like translation and summarization is a lot easier to keep on topic and task though, basically only concerned with expressing things already in the text. free-form conversation has a bunch of things against it in that regard (e.g. humans will try the limits of the thing, it is actually *tuned* to drift away from stuff (or it'd just relentlessly repeat itself and what you say), and generally ingesting text it has generated leaves a lot of room for issues to reinforce themselves)

but, again, i highly doubt it'll be very good at telling the point of the meeting from all other things that go into human communication. many meeting summaries starting with the point "why isn't loving audio working in teams again?!".

Shaggar
Apr 26, 2006
ime teams realtime transcripts work really well for people without accents like myself, but given the way our meetings go i'd be amazed if it could generate enough context for usable summation

distortion park
Apr 25, 2011


https://twitter.com/AlphaSignalAI/status/1630651280019292161?s=20

of the three examples in the first tweet, the first is technically wrong, the second gives an overly precise answer on incomplete information, the third seems right though? seems like not the best choice for your hypetweet (although i respect the paper authors for leading with those examples, if you look in the paper the rest it does better on).

distortion park fucked around with this message at 11:43 on Mar 1, 2023

distortion park
Apr 25, 2011


probably good enough for solving captchas and navigating lovely websites to spam them though

Agile Vector
May 21, 2007

scrum bored



College Slice

distortion park posted:

of the three examples in the first tweet, the first is technically wrong, the second gives an overly precise answer on incomplete information, the third seems right though? seems like not the best choice for your hypetweet (although i respect the paper authors for leading with those examples, if you look in the paper the rest it does better on).

the third looks like a braid, but could be technically a ponytail? seems like a safe answer, but the previous two implies overconfidence being masked by generality

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

Shaggar posted:

ime teams realtime transcripts work really well for people without accents like myself, but given the way our meetings go i'd be amazed if it could generate enough context for usable summation

i think its bytedance maybe? or tencent, i forget which. but it has a realtime transcript feature in their teams clone that works pretty well on both mandarin and english, and can even handle both at the same time. like if your speaking mandarin and switch to english for a word or a few sentences or whatever, it picks up on that and it actually works p good

Pollyanna
Mar 5, 2005

Milk's on them.


fartificial intelligence

echinopsis
Apr 13, 2004

Pollyanna posted:

fartificial intelligence

prolly your best post ngl

Pollyanna
Mar 5, 2005

Milk's on them.


good

Pollyanna fucked around with this message at 20:20 on Mar 1, 2023

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

hows gpt 4 these days

fart simpson
Jul 2, 2005

DEATH TO AMERICA
:xickos:

does chatgpt have a soul?

echinopsis
Apr 13, 2004

yeah but itís a broken one that requires salvation

rotor
Jun 11, 2001

classic case of pineapple derangement syndrome

fart simpson posted:

does chatgpt have a soul?

no one ask the mormons about this

cowboy beepboop
Feb 24, 2001


i don't really have any questions for any AI

Shaggar
Apr 26, 2006
https://www.youtube.com/watch?v=Ktg8E7i4nzw

post hole digger
Mar 21, 2011
ive been playing with chatgpt and gpt-4 more and its not hard to think that its going to ultimately replace traditional web search to a large degree, and that will basically be the culmination of the closed web project that started in the social media era.

it has a long way to go to that point, and i'm not really convinced its going to take that many people's jobs, because the value of the results you get is directly proportional to what you are asking it, and in how much detail, not to mention being able to know whats wrong with/how to improve the results you get back. but i think part of what impresses people so much about chatgpt is you can type in a question and get an answer (even if its wrong, lol) which has become less and less likely using traditional search with the proliferation of seo garbage and constant algorithm tweaks that have all but ruined it.

echinopsis
Apr 13, 2004

if you can imagine the kind of contagious amazing personality that a cult leader must have to make people follow them

will ai be able to be that good

echinopsis
Apr 13, 2004

imagine if ai learned how to hypnotise

akadajet
Sep 14, 2003

ai should be like the computer from star trek. it can just do whatever the plot needs it to do for that episode

nudgenudgetilt
Mar 18, 2003

it scratches that "oddly satisfying" itch

akadajet posted:

ai should be like the computer from star trek. it can just do whatever the plot needs it to do for that episode

isn't that what we have today? ai can do whatever it takes to raise vc

Agile Vector
May 21, 2007

scrum bored



College Slice

nudgenudgetilt posted:

isn't that what we have today? ai can do whatever it takes to raise vc

feregiPT

Adbot
ADBOT LOVES YOU

Agile Vector
May 21, 2007

scrum bored



College Slice
like elon, it too focuses on the importance of :females:

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply