The first thing to do in this article is settle your nerves.
I’d guess half the readership are excited about the potential for machine learning and what that might mean for the future of business.
The other half will be mentally storing weapons to halt the onslaught of robot overlords coming to threaten our very existence.
So let’s start there.
This article isn’t to discuss the latter. It’s about harnessing the incredible power machines can give us to help us make better decisions and guide supply chain strategy.
A fair while ago, it was widely accepted a computer would never beat a human at chess.
The computations needed to do so were far too complex. You see, it’s not about having a simple strategy. It’s about that strategy changing and making decisions based on an exterior force altering the landscape.
In this case, a competitor who makes decisions of their own and changes the way their opponent should think.
Sadly, a computer beat world chess champion Garri Kasporov in 1997. So that plan’s long gone. The next hope for mankind was pinned on the ancient Chinese game Go.
Where the complexity of Chess was out-manoeuvred in the 1990’s, Go has a much more complicated board of 19×19 squares, and therefore needs far larger calculations for each move.
In fact, after every single move in Go, it’s estimated there’s 3,000 possible combinations to rebuff your opponent.
So there’s no way a robot could beat a human there, right?
Sadly, wrong again.
In fact it happened 4 years ago in 2017. A Google A.I. program claimed victory in the first of three matches.
“Last year, it was still quite humanlike when it played,” Go champion 19-year-old Ke Jie claimed at the time. “But this year, it became like a god of Go.”
So, it’s learning. And was learning as it played in real-time, which is one of the fundamentals for A.I. programs.
So how can this be good for business?
And specifically your business?
Well using A.I. programs for the complexity of supply chains has actually been going for quite some time.
Automation of processes in factories is where the biggest use case of A.I. was seen. But now, developing machines to learn problem-solving by making the best decisions autonomously, is where the focus lies.
What exactly are we talking about here?
OK, let’s break down the basics of three different elements of our robotic future.
First of all, we have A.I.
A.I. or Artificial Intelligence is the science of programming machines to mirror human (or animal) behaviour.
For this to happen the robot needs to learn on the fly, something many humans struggle with.
But much like humans, a key observation so far has been that robotic A.I. actually learns quicker when their environment’s harder to navigate. Meaning they learn quicker from mistakes.
There’s a question of an existential nature for us humans there too, but let’s leave that for another time.
An obvious challenge to many businesses is finding ways to collect & process the sheer volume & variety of data available.
Next, there’s machine learning.
To a degree, A.I. can’t exist without machine learning. In fact, it’s the discipline within A.I. dedicated to the study of algorithms.
Simply, the ability to perform a task by analysing data.
In humans, let’s use walking as an example. We’d see a potential slope ahead and alter our bodies, or effort to negate the heightened difficulty of the task.
Robots or machines also need this ability. Should you see a robot make a change to its behaviour to overcome a change in exterior conditions, they’ll be collecting and storing a huge amount of data (what we’d call ‘Big Data’) to perform that alteration.
Robots without A.I. programs simply fall over when their exterior landscape changes.
Finally, Analysis and Data Science
Data Scientists are a rare breed. They’re a cross between a Scientist and a Mathematician in that they take huge amounts of raw data and give it a story. They can find a plot from raw numbers.
These days many businesses collect monumental amounts of data. But turning that into useful information can be difficult. Let alone a story which can guide business decisions.
And so Data Science is the art of finding a storyline in the analysis of data, which can help businesses make decisions about the future.
Which meanders nicely back to the question in the title… Can machines make good decisions for your business?
And to answer that question, let’s look at how machine learning is already benefiting supply chain teams.
The current situation
Let’s take inventory management as an example. Tasks like building forecasts, planning promotions & working with suppliers were always limited by the data available and the rules humans put in place.
But this way of working now looks archaic compared to what machine learning can enable businesses to do.
Examples of machine learning in action
With demand forecasting, the guesswork’s eliminated. Machines can now review historic sales and autonomously ignore anomalies, for example eliminating the data from panic buying toilet paper so as not to soil the picture of future demand.
But it doesn’t end there, businesses can now capture external data such as weather & social trends to perform even deeper analysis.
The machine instantly processes all of this data to autonomously determine optimal order levels to hit future demand, at the lowest possible cost.
Essentially machines are able to do far more of the heavy lifting.
And where human intervention is required, supply chain teams have richer insights than ever to make better decisions.
Better science means a better business
Allowing machine learning, A.I. and Data Science to guide your supply chain can improve your business in a hugely exciting way.
But it’s crucial to realise the main objective of this strategy, and that’s improving the business as a whole. Not just one department at the cost of others.
To truly change the game, you need to bear in mind sharing information between departments is absolutely critical. Of course it is already simply dealing with humans, but remember machines have less ability to infer things.
Equally as with any data, quality and reliability guides your decisions. So incomplete data sets or a lack of information can limit your success.
But under the right conditions, and with all the right information at your disposal, the future’s incredibly bright for those bold enough to harness the power of machines.
