By 2020, 20% of companies will dedicate workers to monitor and guide neural networks. Such workers will not think or excel in the same ways your best developers work today
What makes artificial intelligence remarkable is that its performance improves as it ingests and acts on data and content. What makes that challenging is that it demands new kinds of operators and administrators.
Artificial intelligence (AI) needs people who tinker with data, who tinker with rules, who tinker with measurements, and who can do things others are not comfortable doing.
We need people – you need people – who are eager to tinker with artificial intelligence to fuel new insights for organizations. People who can tinker with flows of big and bigger data the way that a chemist tinkers with the fuel that allows jet engines to rise, or the way a nutritionist tinkers with food to help an animal thrive. We predict that by 2020, 20% of companies will dedicate workers to monitor and guide neural networks. Such workers will not think or excel in the same ways your best developers work today.
Finding data and content tinkers isn’t the same as finding software tinkers. The workers you’ll need to develop your neural networks, and fit them into business strategies, must enjoy the process of working with the new kind of intelligence that artificial intelligence represents.
AI truly is different. Humans think with evolutionarily unprecedented processing power that reveals to us patterns in the world with the ability to feed us, clothe us, and grant us safety in what was only recently (in geological time) a world that was pure danger. Animals have their own smartness (only rarely “intelligence”), which evolution has honed to target their own needs for individual improvement or species survival and expansion.
AI can be directed to just what humans want or need. Many observers note that while AI may be infinitely complex and confusing, it’s humans who develop and improve its power, so it can never be truly “artificial.” I get their point, but that’s a little like saying that a skyscraper is the same as a termite mound in terms of its “naturalness.” It might be true, but saying it blurs boundaries in a way that only a philosopher can truly embrace.
AI demands the guidance of human intercession to manage what data it consumes, and how it should learn from that data. These people are the tinkers we all need to hire, tinkers who understand the impacts of the rules that AI begins with and the way that those rules – and the superstructure of rules that AI builds on top of them – affect the business outcomes that organizations seek.
Such tinkers will craft the tests that examine how the AIs work. They’ll evaluate the results of the tests, and alter the mix of the data, in terms of where it comes from and what is recorded within it, that adds value to the AIs in their charge.
They’ll be hard to find, these tinkers, because many of them don’t know that’s their new craft, to be the intelligent manager of the AIs that organizations will select to augment their human skills. Such workers must be truly curious, passionate about development, and willing to be proven wrong. Among the essential qualities of AI is that it allows us – and forces us – truly to look at data and our assumptions about what it shows in new ways.
To find your tinkers, advertise for people who are curious and excited about the data environments you need to understand. Look for psychologists, and anthropologists, and scientists of all kinds (including data scientists.) Be willing to experiment, and reward the creative. Give them tests that allow them to demonstrate to you (and themselves) how they can think creatively about how systems work now, and should work. And offer them training; a domain expert might be truly skilled in how a business works, but have bailed from math in pre-Calculus in high school. Statistics might look scary to them, but a touch of it can help understanding how the AIs work and “think.”
Clients tell us they find these thinking tinkers in the strangest places. They discover them as expert providers of a service today – the best (or nearly the best) technicians who can articulate how they accomplish what they do. Or they advertise for them by offering problems as an audition to see what the solution looks like. Or they hire people who are fascinated by the new sciences of variations on machine learning and the opportunities it provides.
AI isn’t how we replace humans; it’s how we make the humans better at what they do, To achieve that force multiplier, we need to go beyond the fundamental programming skills that have made the software world what it is now. New expertise (and a lot of work) must be invested to achieve the artificial intelligence that gives humans greater speed, greater flexibility, and greater scope of action. At every stage in development, different human tinkering skills are valued – from rocks and vines and sinews to code in a million years or so. This time, it’s tinkers in intelligence itself.
Whit Andrews, VP Distinguished Analyst, Gartner