Recently I’ve watched a relatively old interview with Jeff where he mentioned that now he would make a very small change to the book. I’m very curious what exactly this change would be? Are there in his opinion more changes that would be needed after all these years?
Related: is there an errata of the book? Or plans for a 2nd edition?
1 Like
No plans for a 2nd edition, but we are working on a free online book called Biological and Machine Intelligence. I don’t think there is an errata for OI, but I’ll ask @jhawkins.
1 Like
A number of years ago I was asked what I would change. At that time I referred to a small reference to Dolphin cortex that I thought was wrong, but later learned was probably not wrong.
Today I would give a different and more substantive answer. Our current research is looking at a major change in HTM theory. In the book I describe that each cortical region learns sequences of patterns (still true), forms stable representations of changing patterns (still true), and passes these stable representations to the next higher region (still true). However I also believed that higher regions learn sequences of sequences leading to more stability as you ascend the hierarchy. My current belief is that is only partially true. There is increased stability but not because of learning sequences of sequences. You can find more info on this elsewhere in the forum, the details are still being worked out.
There is no errata for On Intelligence.
3 Likes
Thanks for your reply and clarifying this, Jeff.