[ home / bans / all ] [ qa / jp ] [ maho ] [ f / ec ] [ b / poll ] [ tv / bann ] [ toggle-new / tab ]

/maho/ - Magical Circuitboards

Advanced technology is indistinguishable from magic

New Reply

Options
Comment
File
Whitelist Token
Spoiler
Password (For file deletion.)
Markup tags exist for bold, itallics, header, spoiler etc. as listed in " [options] > View Formatting "


[Return] [Bottom] [Catalog]

File:[MoyaiSubs] Mewkledreamy -….jpg (498.34 KB,1920x1080)

 No.17

I wonder when the successor of the LLM will come out. People like Lecun are very adamant in disagreeing with companies like OpenAI in that LLMs are the path to "AGI". I tend to agree with the naysayers, as it does seem like it needs an entirely new framework to even approach such territory. Really powerful tools, but people expect too much from them.
It bothers me that we now need to say AGI to mean what AI used to mean. I don't think we actually want AGI, though, if you ask me. Robot maids with free will? Ehhhh, wouldn't you prefer an idealized artificial personality, albeit one with some randomized freedom protocols or something?

 No.20

The idea that not all AIs are the same is pretty old. So I don't see a problem with the denomination of AGI as a lofty ideal.

>Robot maids with free will?
Free will is a bit of a myth, anyway.
You cannot choose your primary wants. If you choose wants, then you do so according to higher-ranking wants. But there is nothing that ranks higher than the highest rank. Those wants are therefore not of your choice. They are what defines you.
If you build a robot maid's AI from scratch, you might be in a position to directly write these primary objectives after which everything else in her personality must follow.

 No.367

Well we probably won't get AGI anytime soon since ChatGPT has entered us into an arms race for the best LLM and that's all anyone is going to chase after for the time being instead of working on other models that could progress us towards that goal of AGI.

 No.368

File:[joseole99][QTS][polished]….jpg (116.4 KB,1280x720)

>>20
Well, in this case I meant something like you could have a robot maid that obeys the command to watch paint dry for 5 weeks straight. The "AI" I'm imagining would have enough knowledge of humanity and the world to find that command ridiculous and she would have the "choice" to reject it, but she doesn't have to understand the deep philosophical implications of why. If she's sentient then she's going to do something else other than being my cute robot maid because that would pretty lame for any intelligent creature instead of idealized artificial personalities.

>>367
Seems like the arms race has been slowing down, though. The money is starting to dry up since all the investors throwing money around wildly without any plan have yet to get their short-term profits and that's all they care about.

 No.369

>>368
It's not that the long-term doesn't matter, it's that exasperated tech guys and business analysts are actively combating the marketing push to have """AI""" everywhere and helping people understand that there aren't actually many good ways to use generative LLMs that aren't just slight improvements on things they already had bots doing, like customer support. The money is in selling the idea of AI, not in anything they will actually accomplish.




[Return] [Top] [Catalog] [Post a Reply]
Delete Post [ ]

[ home / bans / all ] [ qa / jp ] [ maho ] [ f / ec ] [ b / poll ] [ tv / bann ] [ toggle-new / tab ]