DISCLAIMER: This bit of news is based from a post on the forums over at The Verge. It is based entirely on rumors. But it just seems worth sharing.
December of 2011 was a very exciting month for Android – No, not because of the availability of the Galaxy Nexus, or the roll-out of Ice Cream Sandwich. Rather, the rumors of a secret project inside of the secret Google X lab, called Majel: a voice-action technology said to be far more advanced than anything we’ve seen so far.
If a post on Reddit by an alleged former Google employee is to be believed, the artificial intelligence behind Majel is so powerful, that it passes the Turing Test (which rates a machine’s capability to exhibit intelligent behavior) 93% of the time. But to make it work accurately at a personal level, it needs a great degree of data.
Where is this data available? Why, with Google of course! Only issue was, Google’s individual services kept their data about a user independent from other services. As is suggested in the forum post on The Verge, this posed as a hindrance for Majel. It would require the provision of permission to Majel explicitly by the user for each and every service.
Now, I’m not going to say this is the only reason behind the new policy. I was honestly surprised that Google hadn’t brought the personal data together earlier, itself. For a company that makes its revenues by an understanding of our nature and our needs, it seemed foolish to have the data separate for so long.
Yet, despite spending quite some time thinking, I couldn’t come up with any current Google service that would benefit from the new policy to the degree Majel would. Maybe Google has something else up its sleeve. I might help them improve their search algorithm or ads.
But I’d like to dream that it’s about Majel. I personally can’t wait to have such technology available soon.
Si une poster sur Reddit par un employé de Google présumé ancien on en croit, l’intelligence artificielle derrière Majel est si puissant, ce qu ‘il passe l’ Test de Turing (qui évalue la capacité d’une machine à exposer un comportement intelligent) 93% du temps. Mais pour le faire fonctionner correctement à un niveau personnel, il a besoin d’un grand degré de données.
Où est-ce de données disponibles? Pourquoi, avec Google, bien sûr! Le seul problème est, Des services individuels de Google ont gardé leurs données sur un utilisateur indépendant des autres services. Comme il est suggéré dans le post sur le forum sur Le Verge, ce qui posait comme un obstacle pour Majel. Il faudrait la mise à disposition de l’autorisation de Majel explicitement par l’utilisateur pour chacun des services.