4 ms·
I listened to a panel recently discussing similarities and differences between human and machine learning. One thing I found interesting, was a human researche
by jimbokun 17y ago
I listened to a panel recently discussing similarities and differences between human and machine learning. One thing I found interesting, was a human researcher who disputed the commonly held idea that computer memory is perfect and human memory much more faulty.
He pointed out that this comes from comparing human memory to simple RAM or disk access. Given an exact memory address, file path, etc. the computer always gives back exactly the right data. However, maybe Google is a better metaphor. You send in a query term or terms, and then evaluate whether what came back was really what you were looking for.
This highlights the importance of associations, indexing, and everything else that goes into using one piece of information to retrieve another.
Google doesn't always get it right, and there are lots of aspects of human memory that Google can't replicate (yet). But it's interesting that the article concludes with an acknowledgement that Google can remember for us much of the trivia we often forget.
- Hoff 17y agoYou've missed one variation: I've spotted cases where Google "forgot" something. Where Google had the information in part of its index and could retrieve it, but didn't have that same information in its more general searchable index. Cases where you could get the data back from its residence (on Google's own servers), but where Google's indexes didn't offer up the data.
- nopinsight 17y agoThis is like when our brain remembers something but cannot recall it, unless reminded of something similar in either structural or superficial features. Maybe Google (or other search engines) could improve its 'similar pages' functionality by emulating some of the ways brains recall things.