this post was submitted on 03 Sep 2025
28 points (93.8% liked)

Futurology

3254 readers
124 users here now

founded 2 years ago
MODERATORS
voidx
Lugh
Espiritdescali
AwesomeLowlander
28
Scaling hasn’t gotten us to AGI, or 'superintelligence”, let alone AI we could trust. The field is overdue for a rethink. What do we next? (archive.ph)
submitted 5 days ago by Lugh to c/futurology
8 comments fedilink hide all child comments
 

Scaling hasn’t gotten us to AGI, or 'superintelligence”, let alone AI we could trust. The field is overdue for a rethink. What do we next?

top 8 comments
sorted by: hot top controversial new old
[–] MushuChupacabra@lemmy.world 14 points 5 days ago (1 children)

Stop investing, make them pay dearly for the obscene power consumption.

[–] Wigners_friend@piefed.social 9 points 5 days ago

Consequences!? For rich people!? Are you mad!?

[–] happybadger@hexbear.net 5 points 5 days ago

A jihad against thinking machines, then revert the internet to 1997.

[–] anothernobody@lemmy.world 4 points 5 days ago

Put more money into it of course!

[–] scintilla@crust.piefed.social 2 points 5 days ago

Maybe try something else instead of acting like LLMs think anything like humans do.

[–] m532@lemmygrad.ml 1 points 4 days ago

Scaling makes the machines better at what they are doing: working for us. AGI would not make sense for this.

[–] Agent641@lemmy.world 1 points 4 days ago

Give em the nuclear launch codes. They'll show their true colours

[–] Zozano@aussie.zone 1 points 5 days ago* (last edited 5 days ago)

Dude just ask the LLM how to make AI