Page 98 - The EDIT | Q2 2017

98
Thoughtleader
PART ONE:
EXPONENTIAL AMBITIONS
In this first section, we explore the technologies
that will most drive change in our planning product. It’s an obvious but easy mistake to think that
human progress is determined by developments
in technology. It’s not. We progress instead based our ability to leverage innovative technologies and capabilities. Technology is only as good as what we want to do with it — and because the development of technological progress speeds up exponentially, we should have exponential ambition to leverage and harness it... because the technology around the corner is transformational.
It’s the technology, stupid
AI and robotics, mixed reality, voice-recognition and VPAs are the technologies to watch for the next couple of years. Between them, they will transform not just our industry, but huge categories like transportation, finance, professional services and retail.
Arti cial Intelligence
It’s just not possible to talk about up and coming technologies without discussing Artificial Intelligence, which, in many ways, is the arch-tech
— or central technological force — that will shape many of the tech developments around it over the coming decade. It’s been a long time coming, and now it’s almost impossible to read a publication or newsfeed without the topic being discussed.
PHD literally wrote a book on AI — and if you haven’t already (what have you been doing?) you should read Sentience, available in a PHD office
near you. It still serves as a great companion piece to our 2017 publication Merge. As you consider AI right now, I recommend that you consider three questions to get a handle on it: what is it, what will it do, and — crucially — who will own it?
What AI is, is non-human intelligence — where intelligence is perceiving the environment, context
or situation and taking considered actions towards
a goal. To do this requires processing data in context, and then making decisions based on that information. Humans are very good at this in some ways and very bad in others. For example, humans are good at understanding relative changes (the water in that glass is hotter or colder than the
water in that other glass) but terrible at measuring absolute values (the water is 22.3 degrees)... this makes us generally pretty good at things like fast intuition but bad at things like pattern recognition in big data sets and at tasks that take a long time and that require constant concentration — which we find exhausting. We’re better at the creative stuff, right?
Unfortunately, machines are becoming increasingly good at the things we find easy and which used to be considered more ‘human’. For example, in May 2017 after Google Deepmind’s Go-playing algorithm AlphaGo beat world Champion Ke Jie, the player said that the algorithm was “like the ‘Go god’... it
is brilliant”, students of the game described the machine’s play as innovative and beautiful.
AI machines are already better than humans in most respects when it comes to narrow AI, performing a single task or operation well — like playing Go, or driving cars. These single-task functions are where we will see the greatest immediate disruption to
our product. For example, planning programmatic media investment requires crunching multiple data sets very quickly and making decisions on where to bid on and place impressions. Using reinforcement learning, AI machines are already better at
‘understanding’ which impressions are working best (attribution) and then optimising media by buying more of these at the best price.
The next wave of AI disruption will come when AIs develop strong, or ‘general’, levels of intelligence — basically the intellectual capability of a human. This
THE EDIT ISSUE 06 | Q2 2017