The world, or rather the devices within it, are getting smarter by the day. This is particularly the case where artificial intelligence is concerned, which seems to be more widely applicable and functional than it ever has been, even though it is still in its fledgling stages.
In spite of our limited understanding of human consciousness, we have simulated it to some extent in the various types of artificial intelligences currently under development across a number of industries. Smartphones come with virtual assistance, companies can process data with accuracy and speeds never before possible, algorithms can guess what we are thinking or offer us semi-engaging conversations. They can learn from us, chat to us, outperform us, improve on us, and in some cases, perform some of our most complicated tasks with comparative ease.
In some cases they promise to provide simple solutions to many of the concerns of our way of life, often trumping what might have been an amazing form of innovation, had AI never arrived to steal its thunder.
This was the case when Huawei showcased the incredible power of their Mate 10’s on-board AI earlier this year, in a demonstration which left the concept of a self-driving car a little underwhelming in comparison; but we will get to that in a moment.
A Little More on Huawei’s Dedicated AI Chip
The Mate 10’s AI chip, otherwise known as the NPU (Neural Network Processing Unit) makes the phone a first of its kind, affording it incredible and intuitive functionality, designed, as all of these products are, to make our lives a little easier.
It allows for machine learning and AI calculation functionality on the device, improving its user experience, and possibly opening the door to making our smart-phones quite a bit smarter than we have ever imagined.
The most heralded function of the NPU is its live scene and object recognition camera feature which has the ability to recognise objects in the real world, and apply a pre-programmed action in response to it. This could be a face, an object, a location or something as simple as the type of lighting available to a scene.
Used in its simplest form, the application here could help users take better photos with less human calculation and work, allowing the device to do all of the calculations of light, focus and distance for you. In our selfie-filled, profile-pictured world, this is a genuinely useful application of an artificial intelligence, but it is not even nearly the most practical use of the tech.
We May Finally Have a Babel-Fish
AIs that can perform instantaneous translations of a text are nothing new, though they still have a long way to go. This doesn’t make them any less impressive, however, or any less useful.
They represent the first step in developing a truly global community, where real-time communication across language barriers can happen seamlessly, if not with a modest amount of confusion and humour.
These translation tools, fuelled by AIs, generally get the gist of it however, only making mistakes where nuances in language and cultural experience make up the meaning of a phrase, such as in the case of metaphors and idioms; though even these are getting better and more accurate by the day.
The Mate 10, or rather its NPU, is able to perform real-time translations of a text by simply pointing the camera at it, at which point a translated text would be overlaid on the real one to make it readable in the user’s own language, as crudely demonstrated in the image above.
Understanding the User
AI and machine learning have been used extensively to understand the habits of people, and in the case of the Mate 10, it is particularly focused on learning more about how users operate the device. It builds up an understanding over time of aspects like the user’s reading speed and vision strength and which apps are important to them.
By learning this information over time, the device can optimise itself to fit the preferences of the user, without them needing to toggle a single setting.
While this type of artificial intelligence is nothing new, it is the first time that it has been made available, with its own dedicated architecture, to a mobile device, and as such has had some powerful implications for the Mate 10’s user experience.
By intuitively reacting to the way that you use your phone, the Mate 10 can automatically customise it to give you the most comfortable experience, depending on how you use it, how often you use it, and what exactly you use it most for.
In short, it learns your habits and adapts, making each experience with the Mate 10 a unique one.
A Particularly Smart Power Plan
This allows the NPU to take steps towards solving a common problem in the 21st Century, that of managing power-plans and battery life on mobile devices.
The intuitive processes of the NPU allows the device to, once it has learnt your habits, prioritise those processes on your phone which are most important to you.
By finding out which apps and functions are most used, the device can allocate a certain amount of resources to those tasks while taking away from those that are of a lower priority.
If you use the device for mobile gaming, for instance, the AI will store a certain percentage of power so that it can be available for gaming, allowing the screen to be brighter or the volume to be higher when these apps are used.
This makes the problem of monitoring and managing battery power a fairly simple and streamlined process, allowing users to make the most of available power when using the device.
Being Driven Forward with AI
Now we move on to what I consider to be the most impressive function of the Mate 10 (or rather its NPU), which has a lot to do with its object and environment recognition features. This isn’t exactly available for commercial use (since it is probably still quite complicated and dangerous to get right), but is still an astounding piece of insight into the possible applications of the recognition features.
Do you remember all of the excitement and buzz around driverless cars that suddenly died down? Well Huawei has just made the concept of a specially designed AI driven car redundant. Yes… redundant.
The idea was promising, don’t get me wrong, but the average car owner showed very little excitement at the prospect of dashing out and getting one. Then there was talk of their communal applications which seemed to just taper off somewhere; but all of that is history now, in my opinion, and we have Huawei’s NPU to thank for that.
In a bid to test and demonstrate the power of its recognition features, the Huawei team managed to teach the AI how to drive (yes, drive) a car. It may not be as proficient at it as humans are, but the fact that it can perform pre-programmed functions in operating the car as a response to changes in the environment, is an eerily impressive development.
This could have far bigger implications for the motor industry than any driverless car could, since it still gives the car-owner the option of taking the wheel. It would, in the ideal world, be compatible with any car that the phone is fitted to, allowing it to operate it without intervention from a human driver.
It will likely be some time, however, before this becomes a commonplace part of national roads, but the potential is there thanks to the NPU.
On-board and Always Available
This is all very impressive, but the cold truth is that none of these features really offer anything that an AI couldn’t already do. So what is it exactly that makes the Mate 10 so special?
It represents a shift in the way that AI capability is being processed and delivered to us. Generally speaking, devices that use AI draw from resources available over the cloud, meaning that none of the work is actually done by the device, but is simply displayed on it.
The Mate 10 is the first (in probably a long line to come) of smartphones that are actually infused with artificial intelligence, doing all of the work itself.
Where other devices would need an internet connection to use these functions, the Mate 10 doesn’t.
Implications for the Future
If the advantages of this type of architecture are evident to users, they are obvious to Huawei’s competitors. In having set a new standard for AI functionality, we will likely see Huawei pioneering a movement which will quickly be adopted by other manufacturers, across a range of devices and appliances.
Consider the difference that smart-processors once made to the mobile-phone scene. Starting from one impressive product and eventually permeating all of them to become a commonplace part of our lives. Through Huawei, we are now seeing a new type of processing come to the forefront which will likely become a standard soon.
Consider the power this will give devices which we already consider to be smart. Imagine a fridge which monitors your nutrition intake or handles everything for you when you need to restock. An ear-piece that analyses speech and translates it in real-time. A security system which can recognise individuals and is able to class them as a threat, a guest or a family member… the possibilities are practically endless, and something tells me that this is just the tip of the ice-burg.
All of this could be available with its own built-in AI, no longer being reliant on an internet connection or central computer.
Contact Applord to Take Advantage of the Power of Mobile
These types of developments are a cause for excitement for our developers. They offer us new tools that can be leveraged to provide your audience with the type of engaging content that they are looking to interact with. In time, we hope to see this tech grow and become widespread, and we aim to be at the front of these developments, making the most effective use out of them. We are a mobile application development company, and as such, are in a position to engage these types of developments to benefit you. Contact Applord to learn more about our comprehensive and customisable development services.