Featured | June 15, 2017
4 Things We Learned From Google’s I/O Conference
Google’s annual IO conference took place last month and, since we’re a conscientious bunch who are well aware of hectic schedules, we’ve compiled a lovely Google IO review just for you.
1. Artificial Intelligence Takes Centre Stage
— Becoming Human (@BecomingHumanAI) 8 June 2017
AI was a focal point of this year’s conference, with CEO Sundari Pichai making it the central subject of his keynote speech.
With Google looking to re-architect the company to be more AI focused, it appears as though the search engine will venture further than their current undertakings which already include Google Photos and Google Drive.
Alongside this, the launch of its new machine learning research project was announced.
AutoML aims to improve its current machine learning systems to the point where it can create its own.
As it stands, building the models is a time consuming undertaking that is pretty darn expensive too (not that that’s much of an issue for $100bn company mind).
If machine learning could be improved to the point where it basically builds itself, it would mean significantly faster development times, alongside a much larger scope for future projects.
A bit similar to a self-cleaning oven really.
2. Google Assistant Owns iPhone
— Google (@Google) 17 May 2017
In short, Google Assistant has wiped the floor (ish) with Apple’s Siri due to the huge steps it’s made in the world of machine learning.
And it’s partly due to these leaps in AI that Google Assistant will now be available on iPhone.
Cementing itself as an intrinsic part of Apple’s future iOS updates, Google will hope to see an increase in voice related searches from users of all devices.
Fingers crossed it’ll start to understand my thick Welsh accent too.
3. Some Interesting Infrastructure Developments
— John Kelleher (@johnkelleher) 1 June 2017
Last year, Google’s Tension Processing Unit (TPU) promised to be a game a changer, and this year, they’ve improved it once again.
The second generation TPU looks to speed up machine learning, how fast you ask? 180 teraflops, that’s how fast.
Side note: If, like me, you weren’t sure how fast that is, I can honestly tell you that it is very fast.
This development to the core of Google’s AI dreams will bring them closer to their goal of creating machine learning programmes that build themselves.
All pretty impressive if you ask me.
4. Bringing the Job Market to the 21st Century
— MO (Markets Online) (@Markets_Online) 29 May 2017
Google for Jobs looks to use artificial intelligence to predict what users will be searching for in the marketplace, and link them with employers who match their specific skill set.
With many platforms hoping to perfectly predict the searching habits of us humans, it appears as though Google is attempting to narrow the field slightly and branch out into HR along the way.
Hopefully I won’t be needing it anytime soon, but it’s always good to know it’s there.