@dark_coderz GPUs have multi DOT product cores to do matrix math in a single cycle across 26 or more floating point numbers. CPUs have more general instructions and they are good at moving memory between devices and memory.
Maybe there are three goals: find errors, fix them if possible, and find places where human intelligence missed the obvious (low hanging fruit) then where humans missed something big. Human intelligence is limited by its very nature but AI is different. We hope it reveals great secrets we missed.
These are incredible. It seems that the surface of mars is too harsh for humans. It’s like it maybe a best practice to have Optimus robots with Boring technology build underground bases with all the radiation blocking from the rocks above the habitats. But the mall and living space you depict is probably how it will appear.
@qianshanmux1 Chip manufacturing is an example of the distribution of wealth through the integration of different specialties across the entire world. It’s the private sector enabling financial success across international borders.
@PhysInHistory I have met scientists that were very arrogant and they believed we have solved everything and there is nothing more. Arrogance isn’t about one world view but instead ingrained in the person.
@elonmusk We reached the asymptote in the Industrial Revolution with factories and the computer asymptote with shrinking gate sizes. Now we must discover the asymptote in AI agents swarm processes. At the moment it might remove the need for government.
@HovrPadPro@luckeyfaraday It seems that language is a small number of rules like use transition words between ideas and the associations between words in a sentence or paragraph.
@luckeyfaraday It’s possible that intelligence is knowledge coupled with pattern recognition. Associative processing like using the DOT product to compute the COSINE between words and basic ideas may be enough to create AGI.
@KatieMiller An “Atlas Shrugged” moment is coming in Texas and the federal government. Does the US continue to be a leader in AI or do we regulate ourselves out of the market?
@Semiconsight The Nvidia board underestimates the worldwide influence of a great leader like Jensen. His connections and 30 years of collaborative work are the life blood of great leadership. They will regret this recent decision.
@brenthuff11@grok It’s because of CUDA. Proprietary languages drove developers to the Nvidia platform. It is also because Jensen listened to his engineers and drove the company towards a great platforms.
@Big_Picture_89 It’s quite telling that the private sector created reusable rockets at incredible expense and challenge whereas government agencies did not. Innovation resides outside the government since private and public sectors have dramatically different goals.
We can think of a token and its vector association as traversing a graph path through the parameters. This is neuro-symbolic-graphs going through parameters. Being able to create discernible speech from the pathway is more important than the value of the parameters showing how the latent space effects the output probability.
@SawyerMerritt The people of Texas benefit from Tesla investments. We are benefiting from in the long run but in the short term Texas gains the wealth of “atlas shrugged” type innovation.
@ytk_matsuda Our shared future as a species to pass the AI singularity into true abundance, removing the need for government and replacing it with individual liberty, free will and the complete removal of scarcity.
@ytk_matsuda Just a note the body is temporary. All the molecules are designed to be recycled. Given the obvious short term nature of our bodies we can accept a “get it done; take action” lifestyle because we only have now.