Machine Learning is not the job killer you fear


Machine Learning is not the job killer you fear

Full disclosure: I am a fan of destroying some jobs. But the public commentary about machines making you redundant is wrong.

The fact is, there are some really boring, repetitive jobs out there and the very smart people who work for you are above them.

Meanwhile, over the last couple of years, we have read repeatedly that advances in data science will threaten all of our jobs. Even lawyers and doctors have nowhere to hide.

I for one, don’t think it will go down like that.

The greatest assistance that machine learning and automation in general is offering us right now, is the capacity to do jobs that we are not even doing at the moment.

Government and private organisations alike either collect, or could collect, vast quantities of data about their operations, customers or constituents. Sometimes we are worried about this, other times we don’t know it is happening or we willingly consent to the process. There is one simple truth I have seen over the last few years that is common to all situations:

It is unlikely that anyone is doing anything with that data.*

At a recent conference I heard organisations talk at length about the immense volumes of data they had, very legitimately, gathered from citizens. After almost half an hour hearing about the types, amounts and structures of the data collected someone (mercifully) asked: “what are the citizen problems that you are using this data to solve?”

This prompted 2 responses:

  1. We are trying to work out which problems to solve…

  2. We are not doing anything with the data. In fact: “we estimate we might be using one 3% of the data we collect, at most.”

The comfortable fact in my mind is that most organisations are in this position. Using data is a matter of connecting with customers, managers and citizens to understand where their issues are AND then imagining what is possible next.

For now, we are probably going to use our data to do jobs that no one does at the moment, because no one wants to do them. Jobs like sorting through millions of words of correspondence to understand citizen or customer sentiment or looking for patterns in hundreds of thousands of sales to dispel our personal theories on when clients leave and why.

So, we have a long way to go before we have to worry about the killer robots coming for our boring jobs. There is just too much low hanging fruit to pick before then.


*Unless they are Apple, Amazon, Uber, AirBnB, Google or Microsoft. In which case, they totally are.

Don SharplesComment