Rachel Coyne on 19 Jan 2018

Machine learning – can the boring bits of your job be replaced by an algorithm?

A machine learning my likes before me?

The other day I watched the ’80s movie, “Pretty in Pink” with my teen daughter. It’s not a movie about machine learning software, but it is a classic. My all-time favourite scene takes place in the “Trax” record store when Duckie serenades Andie by lip-synching to Otis Redding. My daughter wasn’t into it. But she was fascinated by the store selling vinyl records.

Pretty in Pink

“Pretty in Pink” (1986)

Granted the film’s 30 years old, but my young teen didn’t know about vinyl record shops and will probably never have a job working in one. Conversely, I wondered how my own teen brain would have processed the concept of Spotify and machine learning algorithms learning my music preferences and finding music I liked before I did.

An algorithm that learns based on past behaviours

The science of machine learning gives computers the ability to learn and improve from experience without being programmed. Sounds like the stuff of science fiction and can evoke images of robots and cyborgs. But this type of artificial intelligence is very real and has been around for a while. If you use a smart device, social media, stream movies or music, or shop online from the likes of Amazon, you’ve experienced machine learning.

Machine learning software personalised advertising on Facebook

Personalised advertising on Facebook

How does Facebook know I need to buy dog food for my pug? Really, how does it know? Sophisticated machine learning algorithms know. Facebook uses them to continuously trawl through data that we have created from everything we have shared, liked and posted. It learns about our online persona and makes personalised suggestions and adjusts our online experience based on our past behaviours.



Data-centric world & extraordinary cloud computing power

Today there are factors at play that are seeing machine learning becoming more commercialised and moving beyond our smart TVs and mobile phones. Factors driving the rise of machine learning in business are: 1) enormously increased data; 2) significantly improved algorithms; and 3) cheaper and ready access to powerful cloud computing.

It’s now feasible to build systems that learn how to perform tasks on their own without costing a fortune or being restricted to universities or top-tier technology companies.

So, what does this mean for regular business and how might it affect our jobs and careers? The possibilities are endless. But, today we are seeing some interesting applications of the technology in areas that would not have been top of mind.

When we see headlines such as, “Australians need to wake up to the robot threat, with five million jobs at risk” or “Artificial intelligence and automation: could a robot do your job?”, we generally relate the threat to simple physical and manual work. But white-collar workers should think again. Machine learning software can automate repetitive tasks not only in the factory but in a lawyer or accountant’s office too.

Remember I said that 30 years ago (or even 10 years ago) I wouldn’t have been able to get my head around Spotify? Well, today it’s the largest on-demand music service in the world and machine learning is driving its success.

Spotify has over 140 million active users, listening to music every minute of the day. Collecting every single data point, Spotify uses that information to train algorithms and machines to listen to music and extrapolate insights. It’s not a random selection that generates our personalised music recommendations. It is data science and machine learning.

And it’s not just music and movie streaming companies using machine learning to create value and deliver a better service. Any recurrent process with a high volume of data is ripe for machine learning software to automate repetitive tasks for higher levels of accuracy and in a fraction of the time.

What jobs can machine learning software replace?

The largest bank in the US uses machine learning software to complete in seconds what a bunch of lawyers and loan officers would spend around 360,000 hours a year doing. JPMorgan Chase & Co. has developed machine learning software, which reviews tens of thousands of new loan documents a year and has dramatically reduced loan-servicing mistakes stemming from human error. Putting this into sharp perspective, this machine learning software completes more than 41 years of non-stop legal work in a few seconds.

Some may think this sounds apocalyptic and we’re all going to lose our jobs. But, really it redefines what is possible and opens a world of new opportunities. Advancements in technology caused the demise of vinyl records. But machine learning and artificial intelligence have brought us new opportunities and experiences for discovering and listening to music. And using machine learning software to automate repetitive tasks has huge productivity benefits. Like automating mind-numbing, labour intensive contract document reviews frees up time for lawyers to skill up and be productive in other areas.

Machine learning software automate repetitive tasks

Natural language processing and machine learning to automate review of high volume documentation

One of the big four accounting firms, Deloitte, is embracing machine learning to automate repetitive tasks like data processing, anomaly detection, and natural language processing for reviewing massive volumes of contracts. EY uses machine learning to review leasing accounting standards and automating the process of reviewing tens of thousands of leases against new regulations. EY claims its machine learning software reached break-even ROI in less than a year.

Automate repetitive tasks with a disruptive technology

Any company, regardless of industry, will have plenty of processes that could be partially or fully automated by machine learning software. Machine learning is a disruptive technology and in time, it will change the way most of us do our jobs. But, not every facet of business will benefit from machine learning. The greatest potential is for automating high-volume and repetitive tasks that have complex rules and large amounts of data.

It’s always good to remember that although technology may take jobs away, like selling vinyl records, there will also be new jobs created by technology, just like those created by Spotify.


Curious how machine learning could help your organisation improve business productivity?

Call Satalyst today for a free consultation on 08 9355 2807 or complete the form below and we’ll get in touch with you.