Marriage guidance

It's a bit like falling in and out of love. At first you're infatuated, obsessed even. By the end it's all pain, problems, stress and fatigue. Making AI really work for you isn't as quick or easy as it's made out to be, but don't fall out of love just yet. AI can transform productivity in ways that pay exponential dividends.

Let's look at AI adoption as four distinct pillars or stages. These pillars are born out of much trial and error with AI in many different organisations. Pretty much everyone goes through these stages. I offer this guide as a way of clearing your path and speeding your progress.

The first stage is augmentation. Think falling head over heels. This is great; I'm going to use it for everything. Did you know it could do that? Try this, it's amazing. Cue lots of experimentation.

The second is automation. Like in dating when you've gone Facebook official and introduced them to your mates. This phase is characterised by trying to automate everything because you can, not because you should, and getting everyone else to do the same.

The third is governance. Doubts and difficult questions pop up. Should you be doing this? How well do you really know what you're getting into? Is this the right fit? The right thing to do? Problems arise and your virtual partner is found wanting. You start thinking about how to take control of the situation.

The fourth and final phase is guard. You understand what the risks are from using AI and from those using AI against you, and you have put safeguards in place.

Play with it. Test it. Learn from it. But don't let it loose yet.

I won't deny you the fun of the infatuation stage. Play with it. Find things that AI is good at doing, test it and learn from it, but don't let it loose yet.

Before you get everybody to use AI to do everything all the time, take a step back and ask yourself what you should be using AI for and, crucially, what you shouldn't. Reflect on the issues in advance. Do your customers know you're implementing this technology? Would they be comfortable with it? What are you comfortable using it for? How will you ensure your data is protected?

Look at the AI choices available, decide which you'll sign up to across the business, and introduce your staff to it – we've covered before why letting people use free AI on an ad hoc basis poses serious security risks. Make it clear to your staff and your customers what you are using AI for.

Then set it to work and be prepared for there to be hiccups along the way (you're in stage two). AI will hallucinate sometimes. It will get things wrong. But you and your staff will catch them because you have oversight.

If you've taken time to think things through in the early stages, the third and fourth pillars along this path are largely in place. You anticipate potential difficulties so can work quickly to mitigate them if they arise; you've set up a framework for what, how, when and by whom. You've got your eye on security. You can refine your usage and your policies around it through experience.

There are lessons to be drawn from each stage and there's an argument for going through all of them for the experience. Or you can learn from others and arrive more swiftly and painlessly at the point where AI is delivering valuable efficiencies and freeing up human time and energy for other things.