Day one at Google I/O certainly didn’t disappoint! The keynote opened the festival with a focus on “Mobile first to AI first” and Google’s approach to dominate the AI space. This remained a theme throughout the first thought-provoking day here at #io17 and therefore the focus of my opening summary.

There are two sides to this; on one hand delivering data centers capable of handling the vast compute requirements achieved by iterating their TensorFlow processors, on the other delivering consumer features backed by an expanding ecosystem and set of SDKs.

In terms of TensorFlow, the new TPUs will support both learning (the most intensive aspect) and also inferrence (using the trained model) as currently supported. The performance is impressive, a single TPU 180 teraflops, and with built-in high-speed networking a hardware ‘POD’ can achieve 11.5 petaflops and this will be available to developers.

Consumer features cover many areas, but three which are interesting are Google Lens, improvements to Google Assistant and Visual Positioning System.

Lens is Google’s approach to AR, tied into existing products. It could recognise a flower, decode a poster for a concert allowing you to book tickets, or interpret views on a street showing you restaurants, reviews and contact details. It will be tied into Google Assistant for live use and also into photos.

Google Assitant is getting smarter from improvements to Machine Learning, for example, the recently introduced speaker detection. It will also start to see improvements with 3rd party applications, now available for anyone to develop. Google Assistant will also tie in with your TV or phone, providing the driving instructions you requested or helping you search through what to watch.

Visual Positioning System uses Project Tango and extensive machine learning to build an expected indoor map of a building, for example, a DIY store. From there it would be able to accurately guide you to an item required in store. Tango is also getting smaller to integrate, so will start to become ubiquitous in Android mobiles.

As well as direct the consumer benefits above, it would not be complete without mentioning the good machine learning can do – the simple example mentioned improving breast cancer diagnosis. We are touching the surface of AI and it will change all of our lives.

Look out for tomorrow’s day two summary direct from Google I/O 2017!