At the beginning of this year, NTT DATA’s Technology Foresight 2017 predicted that Artificial Intelligence would be the central technology driving change in 2017. This prediction is already coming true. During the recent Google, Microsoft and Facebook developer conferences a new era was heralded in – a move away from ‘mobile first’ and towards ‘AI-first’.
Here is RMA’s summary of the new AI technologies and features that have been launched by these tech giants.
New AI Technologies:
Google – Cloud TPU
Last year Google launched the first version of its AI chip – Tensor Processing Unit (TPU) – a specially designed chip for machine learning that works on the company’s TensorFlow platform. TPUs are responsible for the recent improvements to Google Translate and Google Photos.
During the keynote presentation at Google I/O 2017, CEO Sundar Pichai revealed the second-generation of the TPU chip – Cloud TPU. The powerful chip has evolved to train neural networks, rather than just run them once they’re trained (like the original TPUs), enabling researchers to run more experiments, and faster. This technology underpins everything from Google Assistant and search to Google Photos and the AlphaGo system.
Google has committed to the TensorFlow Research Cloud programme, which offers up TPU resources to developers who agree to publish their findings and open source their code, as well as Google.ai, which helps researchers access the latest in machine learning.
Facebook – Caffe2
Facebook open sourced its deep-learning software Caffe2, during its F8 developer conference in April. Caffe2 is the software that Facebook developers use to train AI models and build apps. It is focused on deploying machine-learning systems into smartphone applications and onto large-scale clusters and enables developers to create chatbots, connect IoT devices, use machine translation and speech, and image classification algorithms for medical applications. Programmers will be able to use Caffe2 to experiment with different neural network architectures more easily than Facebook’s research orientated PyTorch software.
Microsoft – democratising AI
Microsoft has ambitions to ‘democratise AI’ for all. Uniquely, Microsoft is letting developers customise gesture commands, voice recognition and more, rather than limiting them to settings in ‘off-the-shelf’ AI. It is planning to share the powerful AI supercomputer that they are building, along with the AI capabilities that are infused in Microsoft apps via the cloud.
New AI Features:
These new AI technologies are driving and powering the new intelligent features that Google, Microsoft and Facebook are launching, from intelligent cameras, to augmented reality and bots. I will just talk about two of these AI features – Cameras and Assistants.
AI and Cameras
Google and Facebook are investing in the power of cameras and photos with AI. No doubt we will see Apple do the same and launch new AI driven camera features at the upcoming WWDC event.
During Facebook’s F8 keynote presentation, Zuckerburg declared that the Camera is now the most important thing on the phone. Facebook’s augmented reality Camera Effects platform will soon power new augmented reality experiences inside Facebook. Photo filters, games and your surroundings will soon be awash in playful and informational metadata. Similarly, Snapchat and Instagram are using low-level AI to add goofy filters to your face.
Google’s equivalent application, Google Lens focuses on providing information rather than laughs. It combines Google Assistant with the photos app, to enable smart phone cameras to ‘understand’ what you’re looking at. It will analyse your surroundings and display relevant content on your screen, such as the rating of a restaurant, the identity of the species of a flower or play music from a poster of a concert.
AI and Assistants
Digital assistants, such as Apple’s Siri, Microsoft’s Cortana, Google’s Assistant and Amazon’s Alexa already boast features such as reminding people of appointments and advising users to set out early if the traffic is bad.
Google Assistant is embedded in Android devices including smartphones, watches and Google Home. What makes Google Assistant different now is that it is a standalone app and available on iPhone and iPad. It is becoming more conversational and will use Lens to understand what you see. For example it can translate a foreign language and show you pictures of the translated object, such as the food on a foreign menu. Assistant will also be used by Google Home to direct information to the right connected screen, such as your calendar to your connected TV, or directions to your Google Maps app on your phone.
Facebook announced that its mobile AI assistant M will continue to get smarter over time. It will allow businesses to make automation tools for managing customer interactions and will offer food recommendations when you’re chatting about dinner!
The changes to Microsoft’s Cortana could be the most significant. The virtual assistant has always been good but has not been able to compete with assistants built into phones, such as Siri. As RMA’s VP of Design, George Neill said when Apple introduced Siri onto MacOS last year – people do not like talking to their computers in the office. This is set to change however, with the launch of the new in-home speaker Invoke from Harman Kardon, which is powered by Cortana and is in effect a replica Amazon Echo.
Microsoft is no longer trying to flog the Windows phone that no one wants, but wants to improve the devices you already own with Microsoft services. Cortana is set to be the vehicle to launch Microsoft’s new cross-platform strategy making it easier for people to use Windows on iOS, and Android. Cortana will anticipate your needs by suggesting the recent documents you might want to open, or saving your place on an article you haven’t finished reading. The more Cortana is used, the more it improves and becomes useful to the user.
Why compete in the AI race?
The Tech giants are ploughing their research and resources into AI. I’m sure we will see the same from Apple at next month’s WWDC conference.
However, the business drivers for these companies winning the AI race do seem to vary. Facebook, and Google’s business model is to mine personal information using AI to sell things or gain advertisement revenue. While Microsoft is trying to use AI for businesses to solve their problems and gain revenue through business.
While it could appear that the sky is the limit with ever-evolving AI technology, it is important to keep in mind that with new, powerful technology, comes responsibility. As Microsoft’s CEO, Satya Nadella reminded delegates, it is important to “take accountability for the algorithms we create, the experiences that we create, and ensure that there is more trust in technology with each day.”
Related RMA blog posts: