If AI exceeding human intelligence is dangerous, who would you rather have controlling it: the collective intelligence of all humanity or a bunch of AI vigilantes?
@pmddomingos
-
Tech billionaire doubts AI startup’s path to AGI despite investment
By
–
Real conversation:
Tech billionaire: I've invested $30M in this AI startup. Do you think they'll achieve AGI?
Me: They're a good bunch, but no.
TB: Ah well, maybe they'll pivot to something useful. -
LLMs Lack Both Inductive and Deductive Reasoning Capabilities
By
–
Rather, inductive and deductive capabilities are both important, and LLMs fall short on both.
-
Billionaires’ AGI Race: Status Symbol of the Ultra-Wealthy
By
–
You're not a self-respecting billionaire if you don't have your own startup gunning for near-term AGI.
-
Sutskever’s Company Success Guaranteed Through Unachieved Superintelligence
By
–
Ilya Sutskever's new company is guaranteed to succeed, because superintelligence that is never achieved is guaranteed to be safe.
-
Machine Learning Progresses 100x Faster Than Psychology
By
–
The experimental cycle in machine learning is a day. In psychology it's months. Therefore machine learning progresses 100 times faster than psychology.
-
The Trade-off Between Reliable AI and Generative AI
By
–
We wanted reliable AI, but instead we got generative AI.
-
LLMs Outdated: Introducing MMMM-SAWs Next Generation
By
–
LLMs are yesterday’s news. We do MMMM-SAWs (massive multimodal models modulo the stuff that actually works).
-
Greatest Risk from Superintelligence: System Crash Due to Bug
By
–
The greatest risk from superintelligence is it crashing due to a bug.
-
CGI Remains GPUs’ Primary Application, Now Shifting to AI Demonstrations
By
–
CGI is still GPUs’ killer app, except that now it’s in AI demos.