Rob Xu

6 posts

Rob Xu

Rob Xu

@XuRob

Katılım Ekim 2018
55 Takip Edilen3 Takipçiler
Rob Xu
Rob Xu@XuRob·
@colin_fraser @atomlib How do you think they got it better on charts and text? Just training on a ton of specific examples?
English
0
0
0
10
Rob Xu
Rob Xu@XuRob·
@KaiSunWave @GaryMarcus Presumably true AGI would render these researchers and their new projects obsolete, given the projections of PhD level capabilities. If I was at OpenAi and believed AGI was near, I sure as hell wanna be on the side with AGI, not a side project
English
0
0
1
35
Rob Xu
Rob Xu@XuRob·
@GaryMarcus @Miles_Brundage @GaryMarcus given that the bet is around criteria for AGI, does this mean you’ve accelerated your predictions for when AGI will arrive? I recall previously you had (tentative) later timelines like 2050
English
0
1
2
150
Gary Marcus
Gary Marcus@GaryMarcus·
What about 2035 rather than 2027 for my AI bet with @Miles_Brundage ? • I wouldn’t be shocked if all were solved by 2035 (but would be shocked if all were solved by 2027). • Would be shocked if pure scaling of LLMs got us there. • Would be shocked if we solved all of them without substantial contributions from neurosymbolic AI.
Lord Jacovid Rothfellerschild 🌎@JacovidLord

@GaryMarcus @peterwildeford @Miles_Brundage @elonmusk @GaryMarcus, what about 2035 instead of 2027? Would you change anything?

English
10
3
40
9.8K
Haider.
Haider.@haider1·
Yann leCun — Current AI systems lack real-world understanding, reasoning, and memory. However, within the next decade or two, AI will surpass human intelligence, gaining common sense, and safety guardrails to stay under control while completing tasks.
English
75
40
348
81.5K
Rob Xu
Rob Xu@XuRob·
@MichaelEMann it’s looking like a long shot that they will… hard to see any positive trends here
English
0
0
0
93