Translate to multiple languages

Saturday, December 30, 2017

2017 indicates that 2018 will be the year of AI phones | PCMag India - Artificial intelligence

Sahil Mohan Gupta, Editor, PCMag India inform, "Every major tech headline of 2017 has had one underlying theme — Artificial intelligence."

Photo: PCMag India

Every major tech headline of 2017 has had one underlying theme — Artificial intelligence (AI). It has been all about AI, machine learning or deep learning. In 2017, we saw the rise of GPU makers like Nvidia as their products took the lead in driving the focus towards hardware solutions tailored for AI. Interestingly we also saw some phones that have started to come with dedicated hardware optimised for AI. If 2017 is an indicator, 2018 is likely to be the year of AI of phones. Every year we see one underlying technology come to the fore. For 2017, we could say it was bezel-less screens, like 2016 was the year of dual camera phones and 2018, could very well be the year of phones that tout AI optimised hardware.

Who innovated in 2017  
Right now, there are three to four companies that come to mind. Apple, Google, Huawei and Samsung are the likely contenders who could take the lead in this space. Huawei and Apple design their own processors and have already added dedicated neural processing units in their latest Kirin 970 and A11 processors which are used in phones like the soon to be launched Honor View 10 and iPhone 8 / iPhone X. Google has designed the Pixel Vision Core co-processor which has been activated with the Android Oreo 8.1 update and going by Samsung’s intense focus on Bixby it could be working on something as well. AI services and ambient computing concepts are a big priority at Samsung and there is a high possibility that it could be working on something that augments and enhances Bixby in its Galaxy S9, though Samsung didn’t announce anything specific for AI when it announced the Exynos 9810 processor in November...

Who is likely to lead in 2018 
It goes without saying that Apple is going to be a leader in this space having been the first company to have a neural chip in a phone that’s sold at scale. Apple has been a big proponent of on-device processing for privacy purposes as well. The next generation of the iPhone will obviously be a big deal; however, one shouldn’t discount the current iPhone X, iPhone 8 and iPhone 8 Plus models which all also have this neural engine.

At the high-end of the spectrum, you can expect Google to double down on AI focused hardware. The Pixel Vision Core was its first attempt. For the Pixel 3, there could be something more advanced. Imagine, without even a processor on a phone that supports on-device machine learning Google was able to achieve incredible results on the Pixel 2 especially with the camera! Google was able to use its machine learning prowess to enable a portrait mode using just one camera. But then again, Google is the quintessential AI company. Google also bought HTC’s smartphone team to double down on the Pixel line, so one should expect something crazy AI focussed from Google in the latter half of 2018.

Samsung has been making some big jumps in the AI space. It bought Viv in 2016, which was created by the team behind Apple’s Siri. It then launched Bixby and has been expanding it as a platform. There is no hardware as of now to speak of, but Samsung can or rather should change this. It will be very surprising if they don’t have a mobile co-processor for AI for its phone next year.

Huawei is not be left alone in this space. Along with Apple, it is a leader in this space. It’s first AI focussed device was the Honor Magic which was a China focussed device. This year with the launch of the Kirin 970 processor, Huawei launched the Mate 10 devices this year, which have a huge AI focus. Yes, they may not sell even half as much as the iPhone, but for the Android pack, they have been leading AI implementations from a hardware perspective.
Read more... 

Source: PCMag India


If you enjoyed this post, make sure you subscribe to my Email Updates!

0 comments: