We are on the cusp of a new era in technology where decisions and problem solving will be aided by smart machines that sense, learn, infer, and even think on behalf of humans. Conversational user interfaces, which rely on smart machine technologies to facilitate dialogue between people and bots, are garnering particular attention.
Technological advances in machine learning, algorithms, and artificial intelligence capable of suggesting—and even taking action—on behalf of humans is driving the trend. But integrating smart machines into the marketing organization will require an investment, a willingness to experiment, a moxie to fail fast and learn faster, and the drive to develop a continuous-improvement-oriented culture.
In this piece from Razorfish, we look at marketing functions poised for transformation in the world of smart machines, from product development and sales to customer experience automation and C-suite advice. In the not-too-distant future, CMOs will:
- Learn from Smart Machines, which uncover patterns that can be used to predict behavior
- Draw laser-beam marketing insights from mountains of complex behavioral analysis
- Engage consumers in new and meaningful ways with predictive analytics
Today’s machines are made smart by drawing on neural networks, which have become increasingly deep with recent advances in hardware. Designed to model outcomes based on a broad and complex set of inputs, smart machines outperform humans in several areas, especially those that involve uncovering patterns that can be used to predict behavior. This is where their greatest value lies—in the ability to identify patterns their developers never imagined.
Smart machines include autonomous cars, robots, and other cognitive computing systems able to solve problems and make decisions without human intervention. Machines that “think” like humans will help solve huge problems, from curing cancer to climate change.
Economists call smart machines a general-purpose technology, or GPT. Such technologies impact a society’s economic and social structures with forces so disruptive that it eventually leads to the launch of an entirely new economic era. Examples include the Iron Age, the Steam Age, the Information Age, and now the Connected Age.
In 1955, researchers theorized that the creation of human intelligence could be automated. Today, a perfect storm is converting such theory into practice with powerful hardware, advanced algorithms, and big data.
In hardware, new GPU chips (graphical processing units) are capable of modeling 10,000 times as many artificial neurons as those from 2007. By 2020, deep neural networks (DNNs) will be powered by hardware one thousand times more powerful than those today, enabling smart machines to be trained in minutes and seconds versus hours and days.
Smart machines draw upon deep neural networks, or DNNs (which house text, images, voice, and even video). DNNs are particularly good at weighing and analyzing large streams of data to identify patterns their developers never anticipated. In fact, the more varied the input stream, the better DNNs perform. It is the learning attribute of a smart machine that differentiates it from “unsmart technologies,” which always require specific instructions from a coder to take action.