This is a story about AI.

We saw a car on the way home today. A modern car, one of those ‘safe’ family SUVs. It was just ahead of us going onto the freeway.

Its rear, passenger side tire was nearly flat, and seemed to be getting worse. Nothing we could do about it — I tried flashing my lights a couple of times but they didn’t get the drift. In the end all I could do was let them go and hope that it went down gently and not with a bang at highway speeds.

It started one of those normal conversations — “technology does so much for us now maybe we have forgotten how to adult…”. You know the conversation.

And its true — until very recently I owned a vintage Holden. One of those cars released when the big technology selling point was an AM radio; and air conditioning was achieved by winding down the window and driving faster. Yes, I kept a tire gauge in the glove compartment for regular checks (and often spare oil in the rear, just in case).

Now? I don’t know where that tire gauge is. My new car (and it is shiny and red and lovely) has a little app on the dashboard that shows me a little picture of my little red car, with four wheels and four reassuring little “OK” signs.

If my tire starts to go flat now, my car will literally tell me. That is awesome.

As long as it still gives me a choice.

As long as I can’t be undone by a bug or an edge case in the software that flings my ‘smart’ car into limp home mode because it detected something a little weird.

A data scientist’s dilemma

We can train people, one by one, in good car maintenance and continue to drive dumb cars.

Or we can recognise that most people don’t drive cars for the joy of maintaining them — most people just want to go from A to B where the real thing that interests them lies.

And if we can help that happen as developers, data scientists and AI designers — shouldn’t we?

I can spend my life (and I have spent a great portion of my life) training people to use and analyse data to make better decisions. That’s still important — many problems can be solved with the right graphic or chart in excel.

But I simply can’t train you to build an Augmented Intelligence app to answer your questions, when your training and interest is the job you really love (business manager, policy analyst, whatever). It’s the difference between teaching you to use a tire gauge and teaching you to build an animated car with little OK symbols.

Nothing personal — you are really really smart. We are just playing a different ball game now.

Instead, as developers, we need to cover the edge cases — to build systems that protect and enable. Not systems that block. Systems that work for people, without people having to twist themselves into knots (or get lost in the spreadsheets) trying to come to grips with the technology, or the statistics.

And we, as people, need to advocate and talk about what this means. To know when we want to be able to pull out the tools and do it by hand (or when we want a real person on the other end of the phone).

But it’s OK to say that sometimes, we just want stuff to happen for us — for the information to be served up for us to explore so we can focus on the stuff we really care about.

These are the conversations we need to have as AI designers, data scientists and humans.

What can the tech do for us ?