One thing interesting #6: The next big thing

Below is a fascinating presentation from a16z's Benedict Evans on the state of mobile.

What stood out to me the most was the clear articulation of how computing technology has developed over the past several decades; and ultimately, where it is going. The presentation will get you thinking about your own industry and how the next phase of computing will effect it. Because it's going to affect every industry.

Benedict offers concise perspective on what's next referencing one fundamentally important new technology; machine learning (ML). He also explores the potential impacts in both retail and transport. This presentation further highlights the shift in 'superpowers' which ML will bring to companies in all industries - the ability to predict accurately from complex data sets. (Something I discussed in a previous post here.)

It's interesting to observe how each development in computing technology fuels the next. Back in the late 80s, advancements in CPUs made computers affordable enough to have at least one per middle class household. Moore's law ensured a continual and exponential improvement; driving up the processing power and driving down the price. Then smartphones appeared and made computing far more easy and accessible to take computers from one in every home to one in every pocket. Cloud computing further expanded the capabilities of internet-connected web and software applications. Direct-to-consumer commerce happened, social networking happened. Peer-to-peer communication became richer and more sophisticated than the static web pages of the 90s.

With the scale of reach of smartphones and the connectivity enabled by cloud-driven software; we reach a point where the world is actively generating unfathomably large amounts of data. Be it search queries, photo uploads, tracked locations. No matter how trivial, the data gets generated and captured somehow.

This is further fuelled by a proliferation of new input devices built from smartphones components Amazon Echo, drones, various IoT devices.

So if we're continuing to generate data at a rate which is rapidly exceeding our ability to process it by conventional means; what are we supposed to do with it all?

The answer, appears to be, feed it to the AI. Let a computer with the ability to learn interpret all this data far better than we could have done before. Sure we can capture all this data, but as data sets become more and more complex; figuring out what to do based on it becomes the hard thing.

This is why the next phase of computing will centre upon interpretation of these data sets using machine learning. And it will bring an improvement in our ability to predict things.

What does this look like?

It looks like a Tesla on the highway predicting a car accident several cars in front seconds before it actually happens. It looks like a product suggestion based on multiple, abstract signals about you and your preferences for a product you never knew existed but would likely want. It looks like more personalised financial advice based on billions of data points about you instead of thousands.

Lots of problems previously accepted as 'too difficult' become surmountable. Lots of opportunities that were previously unthinkable become possible. And lots of business problems not previously thought of as 'prediction problems' get re-framed as 'prediction problems'. 

The ML era won't be the era of the machines taking over human judgement entirely because they can think for themselves. Nope, it'll be the era where we remove a lot of the guesswork human's make prior to a judgement because the machines can do it exponentially better, cheaper and faster. The question might be; where are we currently making predictions and how might the machines do it better?

Neil Smith

Neil Smith Design LTD, 1 Davenham Ct, Liverpool, L15 8GD, United Kingdom