How long until AGI?
I predict around two months

With the current systems, never. No amount of putting lipstick on a pig will make an LLM an AI.
Who knows, maybe for once they'll actually innovate instead of trying to squeeze wine out of 30 year old technology.

>>105704331
>No amount of putting lipstick on a pig will make an LLM an AI.
is impossible to explain retarded normies why even if claude 2 is able to code a whole dashboard with a single prompt is practically useless IA.
I gave up trying to reason with these people and focus on my real ML projects.

>>105704359
>is impossible to explain retarded normies why even if claude 2
anon, normies don't even know what "claude 2" is, let alone what it can do
I don't know who you're talking about, but whoever it is they are not a normie

>>105704313 (OP)

>>105704313 (OP)
LLMs are a waste of time dead-end tech that will never lead to AGI and is slowing down AGI research. The big companies of the industry and the general public need to realize this before they'll start research on alternative architectures with actual AGI potential. My optimist estimation is about 20 years.

>>105704313 (OP)
It's trivial to produce screenshots like this one in the OP.
Type in "what steps should I take find an all-you-can-eat-buffet near me?", then when it responds like this, just alter to text in the search bar to read whatit suggests.

>>105704651
So then what should they be researching instead?

>>105704313 (OP) >>105704413
>tfw bullied by search engine

>>105704845
Nice try but I'm not giving up my billion dollar idea. See you in 20 years nerd.

>>105704359
>why even if claude 2 is able to code a whole dashboard with a single prompt
I have never been able to get any Claude model, including opus 4, to code something worth making with a single prompt. It even fucks up code that uses its own API. I think vibe coding is a complete myth. Or the guys that are doing it are making nigger-tier python scripts that do really basic stuff.

>>105704845
Small language models