On Tuesday, David Jones of California filed suit claiming that, according to the LA Times, “Apple oversells Siri’s abilities in advertising and TV commercials”. He joins Frank M. Fazio in rallying against the Siri Beta release, assuming that Siri has risen fully formed from the waves, ready to bring perfect love, hope, and voice assistance to all of the iPhone 4S-owning humanity.
It has not.
Although voice recognition software has been on the scene for many years now, and tools exist on handsets from iOS to Android, what Siri does is actually distinct and special, and deserves some, if you excuse the pun, recognition.
As the new iPad demonstrates, dictation features alone do not equate to Siri. Siri is more than dictation. She offers a digital assistant fully integrated to the internal iOS system. As you can read about in Steve Sande and my book, “Talking to Siri“, because Siri works as an OS-level feature, she can make appointments, create reminders, link to Yelp, query Wolfram Alpha, and more.
Not only is that pretty amazing, but it’s something that must grow and develop as an OS feature set. Each task that Siri accomplishes must be coded, tested, and deployed. This takes time.
Siri also relies on adapting itself to a large and varied community of speakers. Its accuracy depends on several things. First, the user must talk consistently, using a speaking style whose recognition can be automated. Want to fool Siri by mumbling or using odd pronunciations? Guess what, you can. It’s not hard to mess things up. It’s more challenging to work on better recognition instead. Steady well-enunciated speech provides a good starting point.
Second, Siri must learn that speech over time. Users cannot expect Siri to provide superior results until Siri has categorized their speech into any of a number of regional dialects. As you use Siri, Siri grows its database that specifies how you and people with your regional characteristics tend to talk.
That’s a big part of why Siri remains beta. The data and recognition patterns are still being developed over time as the product is rolled out. Siri is still learning how to listen.
I obviously don’t speak for all Siri users in the world (again, pun unintentional) but the Apple commercials seem pretty realistic to me, especially for US speakers with fairly standard accents. These commercials appear to show people who have invested a bit of time learning how to talk to Siri and having Siri learn how to listen back.
Should Apple warn users that results may vary, and should improve over time, and that Siri is meant for a single user per phone? I’m not a lawyer, so I won’t go there. But can Siri do what Apple promises in those shiny commercials? I’d say yes.
How do you feel about Siri? Share your successes and frustrations in the comments.