Artificial intelligence… two words that conjure both fear and excitement. The fear comes from the uncertainty caused by sources as diverse as Hollywood (and its depiction of evil machines bent on world domination) and luminaries such as Stephen Hawking, Elon Musk and Bill Gates who have each warned of the dangers of unchecked machine intelligence. Dangers that include mass unemployment as machines, tirelessly and expertly, conquer both blue-collar and white-collar human labour and the eerie eventuality that machines could not only exceed human intelligence, but also evolve into conscious beings that will challenge the human race for dominance.

But, as any techie can tell you, there is a great deal to be excited about as well. AI has the potential to free humans of menial and demeaning work, allowing us to develop our creativity and supercharge our leisure. With the help of our machine companions, we can solve poverty, cleanse our environment, heal the planet and explore the depths of our oceans and the vastness of our universe. Indeed, there is an argument to be made that, without the aid of machine superintelligence, our own species may well be doomed.

This subject is important enough that speculation about what to do (and not to do) should not be left solely in the hands of the technology elite. After all, as much good as the Microsofts, Amazons, Apples, Googles and Facebooks of the world have achieved, they’ve also left us with a legacy of buggy products, screen addiction, hacked elections, revenge porn, online scams, cyberbullying… you get the picture. It’s up to all of us to ensure that this time we soberly and collectively think our technological future through. AI is not a technology to leave in the hands of the few.

So off we go! This is your primer for what AI is, how to think about it, and how to begin pointing your own business and career toward the future. We begin by differentiating AI from traditional software. As much as software has changed the world, it is an archaic process compared to AI. Comparing software to AI is like comparing communicating by hand chiselling a message on a stone tablet and delivering that tablet via horse-drawn chariot across a Roman road to, say, modern video chat. The former is a gruelling, painstaking slog. The latter is effortless magic.

Software is that painstaking slog. An endeavour undertaken by teams of human engineers working months and years to code an app line-by-line, file-by-file, input-by-input, test-by-test until release, only to then discover that it doesn’t work and the process has to be repeated ad nauseam until, eventually, another team of programmers uses the same gruelling process to build a slightly better app, and so on and so on.

AI is different. It isn’t programmed so much as it is trained. To understand the difference, imagine for a moment you see a cat dart across the street in front of you. How do you know it was a cat? How did you instantly understand that it wasn’t a dog, a camel, or a coyote? Think about it, no one ever programmed you to recognise cats and differentiate them from other, similar creatures. No one gave you the instruction: if the ears look like x and the body moves like y then it is a cat. No, you learned that all on your own by observing cats, dogs, camels and coyotes. Your neurons were presented with lots of training samples, let’s call these samples data, and somehow figured out the differences by configuring neural connections in your brain. Voila! You were trained to recognise cats.

AI works on the same training principle. Give it enough pictures of cats and it learns what a cat looks like without being told details about the ears, nose and eyes. Show it enough games of chess and it learns how to play chess without having the rules explained to it. Expose it to enough human language and it begins to understand how to speak it. Let it drive enough roads and it passes a test for a license. AI learns as humans learn, trial and error, rinse and repeat, time and attention. Which leads us to the differentiator between AI and humans… AI has a lot more time, is a lot better about paying attention and will never get bored, tired, grouchy, depressed or anxious. As long as a problem can be reduced to data, an AI will get really good at solving that problem.

This brings us to the next point: reducing a problem to data. Chances are, your job and your business can be reduced to data, opening the possibility that neither you nor your business have much of a future. A machine capable of recognising plants from weeds will replace gardeners. A machine that has learnt how to file taxes will replace accountants. A machine that has learned how to research laws and case precedence will replace lawyers. Your insistence that you and your business are somehow special will be the very reason you both get replaced by an AI. Someone is going to reduce your job and your business to data. My advice is to be that someone! Begin looking at the problems you face as data. Begin understanding that the decisions you make are simply your natural neural network processing data and recognising patterns. What’s the difference between your natural neural network and the artificial ones that have already learnt to play chess, recognise human language and drive cars? Yours needs sleep and diversion. AI does not.

That’s right, we’re outclassed. For the next decade, those who construct and train the machines will be the ones making enough money to tide them over when the machines take over. Start today, begin by recognising the data you use in your own training and decision-making. Reduce the problems you and your company solve to data. Learn how to use AI models and training schemes. Be the creators of the machines that replace you.

Someone is going to do this. It might as well be you.

James Whittaker is a Distinguished Visiting Scholar at Lancaster University Management School, having previously been a Distinguished Engineer and Technical Evangelist at Microsoft and Engineering Director at Google. docjamesw.com