> “When you try to build a system for understanding natural language, and you don’t have many examples of the kind of understanding you want,” Pereira says, “then you have to prescribe, you have to write—essentially teach it grammar—so that it can do the understanding. That teaching is very laborious.”
> When millions of people begin conversing with Google, through the Assistant, the seas of difficulty suddenly part.
This is why Google has been able to continue to dominate search (IMO). Their original algorithm was radically better than the competitors, which allowed them to start to gather extremely relevant, and hence valuable, data. They have then used that data to make their algorithm better, creating a virtuous cycle.
Additionally, the same reason explains why Siri sucks- Apple might have brilliant researchers working on it, but they don't have the same data that Google has access to, which hamstrings them.
The extreme, non-linear value of the data makes me extremely bullish on Amazon. Alexa/Echo are allowing Amazon to gather an extremely large set of very valuable data that they can use. If they proceed to open up Alexa through an API (which I believe they're planning on doing), that value will continue to skyrocket.
> Additionally, the same reason explains why Siri sucks Apple might have brilliant researchers working on it, but they don't have the same data that Google has access to
Nah, I think it's mostly because Google has hired better talent (and more employees working on ML) and invested more money into research. It's only natural that a team of 300 does more than a team of 30, but I am just guessing. Apple has a web browser and two OS platforms, with cloud services. They have tons of data in the system, just like Google.
I have a nexus 5 running android 4.3. Even without this new assistant, I get all sorts of youtube suggestions that creep me out and make me think I am being profiled on every single detail about my life.
3 comments
[ 5.9 ms ] story [ 23.2 ms ] thread> “When you try to build a system for understanding natural language, and you don’t have many examples of the kind of understanding you want,” Pereira says, “then you have to prescribe, you have to write—essentially teach it grammar—so that it can do the understanding. That teaching is very laborious.”
> When millions of people begin conversing with Google, through the Assistant, the seas of difficulty suddenly part.
This is why Google has been able to continue to dominate search (IMO). Their original algorithm was radically better than the competitors, which allowed them to start to gather extremely relevant, and hence valuable, data. They have then used that data to make their algorithm better, creating a virtuous cycle.
Additionally, the same reason explains why Siri sucks- Apple might have brilliant researchers working on it, but they don't have the same data that Google has access to, which hamstrings them.
The extreme, non-linear value of the data makes me extremely bullish on Amazon. Alexa/Echo are allowing Amazon to gather an extremely large set of very valuable data that they can use. If they proceed to open up Alexa through an API (which I believe they're planning on doing), that value will continue to skyrocket.
Nah, I think it's mostly because Google has hired better talent (and more employees working on ML) and invested more money into research. It's only natural that a team of 300 does more than a team of 30, but I am just guessing. Apple has a web browser and two OS platforms, with cloud services. They have tons of data in the system, just like Google.