3 ms·
The training process is finding local minimums based around an initialization vector of random numbers. 1000s of years of evolution probably mean you were initi
by hannasm 3y ago
The training process is finding local minimums based around an initialization vector of random numbers. 1000s of years of evolution probably mean you were initialized better than a baby AI using a pseudo random number generator.
- ddalex 3y agoWatching my kid learn to speak made it clear to me that we are pre-wired for the acquisition of language - we're not building the structures from scratch, we are learning to put words into the empty spaces. Probably natural evolution at best - no speech, lower chance of gene propagation