![Will the LG G5 have use the new Zeroth processor as well as an unusual design? Image credit: CNET Korea]({"default":{"load":"default","w":"73","h":"41","src":"//img-s-msn-com.akamaized.net/tenant/amp/entityid/BBowDqM.img?h=409&w=728&m=6&q=60&o=f&l=f"},"size3column":{"load":"default","w":"62","h":"35","src":"//img-s-msn-com.akamaized.net/tenant/amp/entityid/BBowDqM.img?h=351&w=624&m=6&q=60&o=f&l=f"},"size2column":{"load":"default","w":"62","h":"35","src":"//img-s-msn-com.akamaized.net/tenant/amp/entityid/BBowDqM.img?h=351&w=624&m=6&q=60&o=f&l=f"}})
Provided by Forbes Will the LG G5 have use the new Zeroth processor as well as an unusual design? Image credit: CNET Korea As Samsung and Qualcomm kiss and makeup, there’s every chance that the latter’s new 820 chipset could be powering the Galaxy S7. This is important for a number of business and technological reasons, as Forbes contributor I an Morris details here.
For consumers, however, there’s one real standout feature that Qualcomm is hoping will give Android the edge over Apple’s iPhone 7: its first ever “cognitive computing” (AI) platform ‘Zeroth’.
In essence, Zeroth is a personal assistant for your smartphone. Now, that’s not a particularly new concept: Siri, Google Now and Cortana are household names for most smartphone owners. But what sets Zeroth apart from the competition is the fact that it’s an active personal assistant rather than a passive one. The difference between the two is that a passive personal assistant (Siri, Cortana and Google Now) require the user to input information to get a result. An active assistant will pre-empt your request and make decisions for you based on previous behavior. It does this by using the sensors in smartphones that are constantly absorbing information.
Whilst that sounds terrifying and there’s a clear data privacy question there (I’ll touch on this later), Qualcomm gave me a fairly pedestrian example: photography. Qualcomm has not built Zeroth to just recognise what type of scenes (landscape, portrait, low-light) are in a picture subject: it can also recognise objects, distinguishing children from adults and cats from dogs. In doing so, Zeroth will index your pictures, making them searchable by keywords. And, with your permission, it can access your contact list and social media, and do facial-recognition on your photo albums.
Outside of photography, Zeroth has advanced handwriting-to-text skills, using the phone’s camera to immediately recognise handwriting and change to accurate text. It will also learn its user’s habits to adjust power management, so programmes only run when you actually use your device.
The concept is solid: improve the overall performance of the device with smarter power management based on user behavior, whilst also improving the user experience by unearthing unused and unknown features (most smartphone owners use very few features ), such as complicated manual camera controls.
But two clear obstacles – application and security – make Zeroth a slightly more complicated proposition.
Application seems like an easier problem for Qualcomm to solve, given its Android market penetration. As Applied Futurist Tom Cheesewright explains.
“The negatives are more about lack of application – is this going to be like Apple’s motion co-processors: API-accessible and widely utilised by app creators because of the scale of deployment of that single platform. Or is this going to be more like the PlayStation 3: slow and complicated to code for and only realising value years after its initial release.
“Given the range of devices of which Qualcomm forms the core, it seems likely that access to it will be widely taken up through a natural extension to Android. At which point I am pretty sure app developers will appreciate the additional resource dedicated to getting to know the device owner and learning appropriate behaviour responses.”
Qualcomm does have form here. It’s not averse to working with third parties to ensure that they get the best out of its processors. I quizzed Qualcomm’s Senior Director of Product Management, Sy Choudhury, on this and he was predictably restrained in his response, but reiterated my thoughts. “We are working with multiple OEMs and developers to make use of Zeroth, with OEM and ISV product announcements slated for later this year.”
Proof will be in the pudding as to whether the tools are made available for developers and if they decide to actually use them, because developer participation really will be the barometer of Zeroth’s success. But the sheer number of Android devices/manufacturers/developers means that Zeroth is unlikely to be ignored. It also gives OEMs a way of distinguishing themselves from Apple as well as the rapidly popular budget Android devices that don’t use the latest chipset.
“While there could be specific use-cases where the data is sent back to the cloud, the majority of use-cases would run and remain local to the device. The beauty of ‘machine-intelligence’ is that while the Zeroth engine is trained ahead of time using servers/cloud, that once it runs on the device, it continues to train itself – i.e. dynamically.”
He continued. “An example of this is what we have demonstrated recently with Zeroth Scene Detect. We have trained the engine to recognize over 50 typical objects such as ‘landscape’, ‘adult’, ‘child’, and ‘soccer ball’. So clearly when used in something like a Picture Gallery app, an end-user can search for and find pictures on their device of (for example) babies. But the key here is that they could then label a subset of these pictures as “My Baby Robert”, and the Zeroth engine would self-train right on the device. This would allow the end-user from that point on to do the more specific search.”
It’s not, however, clear what the exact use-cases of data leaving the device are, but they’re likely to exist, which is a cause for concern. Zeroth in effect builds a deeply personal behavioral profile on you, which is catnip for marketers. The concern is that any app developer that uses Zeroth will also want to make use of the rich data gathered by the platform and monetise. This means more intrusive marketing and little privacy on your most private device.
There might not be as much of a privacy issue here as other services like Google Photos, but we’ll have to wait until the 820 is released. But one thing is clear: Qualcomm has to fight to stay relevant, and fight even harder to be chosen as the processor supplier for modern smartphones. For companies like Sony and LG, fighting against the processor prowess of Samsung, Apple and upcoming Huawei, these extra features offer a way for them to make even better devices. And, perhaps that’s why Samsung is rumoured to be using the 820 – it doesn’t want to have fewer features than its rivals.
