#49 - AGI: Could The End Be Nigh? (With Rosie Campbell)
Increments
The Importance of Existence Proof for Neuroscience
I'm not saying that we will never be able to program it I just don't think that we can program it using existing methods where what I specifically mean is grading descent on training data yeah. There are some philosophical reasons why you can't just get new knowledge out by taking existing examples but again this is a falsifiable prediction because you could easily set up experiments where you see if it produces the next round of inventions. If you're going to produce new knowledge you have to have some notion of truth and I don't think you can get that simply by statistical correlations between particular strings so do your question do I think it's possible absolutely as possible? Do I think any of the existing systems that
Remember Everything You Learn from Podcasts
