Black Friday – USA's biggest sales day – is almost above us, and it's good news if you're in the market for a smartphone. Dealers like Target and Best Buy offer hundreds of dollars outside the purchase of flagship Samsung Galaxy phones. It's a buy-and-get-one-free deal on the LG G7 ThinQ on T-Mobile. And even Apple's 2018 iPhone setup, which was unveiled just two months ago, sees price reductions.
It's almost too much good, especially if you have not committed to a brand or model yet. Conventional wisdom will make you compare display resolutions, processors, accessories, and other hardware components to find the phone that most appeals to you. But this is VentureBeat's AI Channel, we suggest another metric: AI features. In our smartphone purchase guide, we have put the top-of-the-range of the year against each other in a match of AI wits, crowning masters in three categories: the smartphone with the best AI chip, the best AI camera features and the best AI assistant option.
Apple's bleeding smartphone trio – iPhone XS, XS Max and XR – have something in common: A12 Bionic, the 7-nanometer custom system-on-chip like Apple featured as its "strongest ever" designed for smartphones. It has six cores – two performance cores and four core cores – plus a four-core GPU. But without a doubt, the real highlight is the Neutral Motor: an eight-core dedicated machine learning processor.
It's Apple's second generation neural engine – the first comes with the iPhone Xs A11 Bionic chip – and it's a step up from its predecessor in every way.
The Nevrale engine can perform 5 trillion operations per second compared to the last generic silicon 600 billion, and apps created with Core ML 2 – the latest version of Apple's machine learning frame – press it to run up to nine times faster and launch up to 30 percent faster with a tenth of the power. In addition, Apple's AI chip utilizes a smartcompute system that automatically determines whether to run algorithms on A12's primary processor cores, GPU cores, neural motors, or a combination of all three.
On the third-party side of the equation, developers can run code on the neural motor. The San Jose software company Nex Team's basketball game, HomeCourt, prompts it to track and log shots, misses, and a player's location on court in real time.
Core ML 2 Neutral Engine Benefits. Core ML 2, for uninitiated, allows developers to load machine model models on an iPhone or iPad, or convert models from frames such as XGBoost, Keras, LibSVM, scikit leather and Facebook's Caffe and Caffe2. Core ML 2 is 30 percent faster than Core ML, says Apple, thanks to a technique called batch prediction, enabling developers to shrink the size of educated machine learning models by up to 75 percent through quantization.
About: Huawei Mate 20 Pro: Leica Lenser
Image Credit: Paul Sawers / VentureBeat  Apple's latest engine is an impressive chip to be sure. And so is the Neural Processing Unit (NPU) in the Huawei Kirin 980, 7-nanometer system on Mate 20 Pro.
It is the second NPU iteration – the first debuted in September 2017 in Kirin 970 – and it is specifically optimized for the kind of vector mathematics that forms the heart of machine learning models. The Chip's two NPUs (up from one in Kirin 970) can detect up to 4500 images per minute, compared to Qualcomm Snapdragon 845s 2,371 images and A11 Bionic's 1,458. And it boasts superior object detection, real-time imaging and real-time object segmentation, and achieves up to 135 percent better performance in references such as ResNet and Inception v3, while using 88 percent less power than Snapdragon 845.
On Mate 20 Pro, System Level AI running on the NPU, intelligently attacks the GPU clock speed during intense game sessions, minimizing system delays and providing "smoother outdoor gaming experiences" in areas with weak signals. In addition, Huawei says its heterogeneous data structure – HiAI – can automatically distribute voice recognition, natural language processing and data sync across the chip.
NPU also features camera features like AI Color, a Sin City inspired effect that holds a subject in color while everything else in the scene is black and white, and a 3D object search tool – Live Object – which recreates real objects in digital environments. Mate 20 Pro's Animoji-like Live Emoji and 3D Face Unlock lock into NPU for face tracking while the camera's Master AI 2.0 mode allows you to automatically recognize scenes and objects and adjust settings like macro and lens angle. In addition, AI Zoom uses NPU accelerated object tracking to automatically zoom in and out of topics; video bokeh emphasizes the foreground while blurring the background; and highlights generate edited video spotlights around reputable faces.
Third-party developers can print NPU through Huawei HiAI Library and API, and some already have. For example, the Microsoft Translator app uses tasks such as scanning and translation of words into images, and the Prism image editor utilizes it to perform device transfer in seconds.
Smartphone with Best AI Camera Features
Google Pixel 3 and 3 XL ($ 799, $ 899)
As we wrote in our review of Google's Pixel 3, have many of the phone's best features – whether it's predictable battery-saving or the ability to screen calls – get better with AI. And it is particularly clear on the front of the photographer.
Pixel 3 has two selfie cams – both 8MP sensors – and on the back of the phone there is a single camera with 12.2 megapixel sensor. The rear camera is the biggest recipient of Google's truly impressive AI. Pixel 3's Top Shot feature takes up a brast frame before and after pressing the shutter button, and uses AI to select the best image. Photobooth cranes machine learns to take the "best" pictures – ie, minimally unclear and well lit – automatically.
Therefore, despite having a camera compared to the LG V40 and the Huawei P20, each of which has three, Pixel 3 takes some of the best pictures we've seen from a smartphone.
For another example of AI's photo manipulative skill, look no further than Google's own portrait mode, which filters out facial features, clothes and skin in portraits with a machine learning model that detects differences in pixel levels in color and borders. (It works on both front and rear cameras.) Another example is HDR +, Google's internal approach to high dynamic range. It captures between 9 and 15 images in rapid succession, and merge them together to minimize noise, giving them all underexposed smart to keep the colors saturated in low light. And yet another is Super Res Zoom, which combines frames to form a higher resolution image that is "roughly competitive with 2x optical zoom lenses on many other smartphones," according to Google.
And then Pixel 3 is really impressive Night Sight mode. Using algorithmically driven alignment and merging techniques including a modified version of the HDR + stack that detects and rejects misaligned frames and a custom auto white balance model trained to discriminate between a well balanced image and a poorly balanced, it is capable of to reduce noise in images taken in environments of only 0.3-3 lux (the amount of light coming to a surface per unit, measured in lumens per meter square). For reference, 3 lux is similar to a sidewalk switched on by street lamps.
Smartphone with Best AI Assistant Option
Galaxy Note 9 ($ 999)
Bixby, Samsung's home-grown voice assistant, gets a lot of well-deserved fools to do a little worse in some scenarios than Google Assistant and Amazon's Alexa. But for all its shortcomings, it has come a long way since the debut next to the Samsung Galaxy S8 in 2017.
The improved Bixby in Galaxy Note 9 – Bixby 2.0 – Better natural language processing, faster response times and built-in noise reduction technology, and it retains The ability to recognize familiar instructions like "Open Gallery app in split screen view and rotate error images" and "Play videos on a nearby TV." It is also more conversational – when you ask about coming concerts over Thanksgiving weekend, for example, it will remember the date range when looking for tickets in the future. Finally, thanks to future support for new languages, availability of more than 200 markets, and the recently launched Bixby Developer Center, it promises broader app support than ever.
There is more to Bixby than the recognition of the vote, of course. Bixby Home, a "social power for your device", is a dashboard with reminders and social media updates along with cards that can be rejected, fixed or permanently hidden.
Also on tap is Bixby Vision, a Google Lens-like object recognition application that uses integrations with Vivino, Amazon, Adobe, Nordstrom, Sephora, Cover Girl and others (and Samsung's data sharing partnership with FourSquare and Pinterest). It can scan barcodes, turn receipts into readable files, and display relevant product listings, recommend wine, display calorie values of food, and make sure you try on makeup products.
So there you have: four flagship smartphones that make innovative use of AI across three different categories.
Apple's 2018 iPhone setup is far away from winning on the chipset front. The upgraded neural engine, combined with powerful software tools and a thriving developmental system cemented in hardware. That said, Huawei begins to pinch the heels, especially when considering Mate 20 Pros first-party camera features that tap the Kirch 980's enhanced AI chip, such as AI Color and Master AI 2.0.
As impressive as the Mate 20 Pro camera is, it only falls short of Google Pixel 3 in the computational photography category. Pixel 3 has one of the best smartphone cameras we've ever tested, thanks a lot to AI.
Last but not least, it's Samsung Galaxy Note 9, a showcase for the latest version of Samsung's Bixby Assistant. Bixby may not be the most reliable platform on the block, but it has grown considerably more robust in recent months. And, in our knowledge, Bixby Voice is one of the only (if not the only voice assistants who can detect chained commands and interact with app menus and submenus, which makes it easy for hands-free use.  IPhone XS, iPhone XS Max and iPhone XR, Galaxy Note 9, and Google Pixel 3 are available for purchase at carrier stores, Amazon, Best Buy and other major electronics dealers of brick and mortar. Mate 20 Pro, worth noting is not made officially available in the United States – you need to trade with a third party dealer to get your hands on it.