3 ms·
One problem is mistaking statistical analysis for learning, data for knowledge
by worik 6y ago
One problem is mistaking statistical analysis for learning, data for knowledge
- visarga 6y agoIt's only a problem if it doesn't work. I'm happy with imperfect protein folders that beat SOTA by 100% and with DALL.E drawing the radish-in-tutu-walking-a-dog and a harp-snail on request. I'll be happy also with the slightly unexplainable medical diagnosis that still beats the average expert in that field. And getting good unbiased data for these algorithms is going to happen eventually.