Our director of operations, Paul Drake, says that while machine learning is no longer the new technology on the block, its potential is still not being used to its full advantage.
Machine learning – that is, the branch of AI that gives computers the ability to gather, analyse and interpret data without human input – is nothing new, but take-up across many industries is still slow.
While the technology is making inroads into the business world, as companies introduce programs to integrate it into their digital strategies, there is still a long way to go.
This delay is due, in part, to businesses either not knowing about it, not knowing how it can be applied or not having a need for it.
Selling points
Machine learning (ML) is all about probability; by applying clever algorithms to large, good quality data sets, predictions can be made on the likely outcome of a scenario that a human couldn’t feasibly achieve.
It is precisely this large amount of data that makes ML so valuable; the technology can use such data to analyse trends and patterns, learning from them to refine itself continually with no human input – essentially, it becomes fully autonomous.
One area in which ML is already proving its worth is in procurement, sifting through data to wean out potential problem tenders and reduce losses.
The benefit is most keenly felt here, largely due to the sheer volume of data involved.
While a human is unable to process such vast amounts, ML allows software to sift through, spot patterns and act accordingly – saving time, effort and money. But the availability and accessibility of data is increasing to decrease their risk.
However, the advantages don’t stop there. Medium and large-scale businesses, by their very nature, tend to involve a lot of digital data. Furthermore, because of their size, incremental improvements in efficiency can, in practice, have a significant impact on the profit margins.
So, using the procurement example above, the ability to predict potential challenges before they happen can have huge benefits on the ability to generate more profit and reduce the likelihood of a loss – something relevant to businesses of all sizes.
It’s not just a white-collar application either. Take the construction industry for example. While it may not be an obvious candidate – it’s a very hands-on field after all – ML still has scope to save time and cut costs. From identifying site risks from photographs to improving design quality, the potential is there – it just needs to be taken up.
At Sapere, we’re already working with firms to see how ML can boost the tendering process, writing clever algorithms that machines can learn from to identity and predict fraudulent activity.
What now?
One thing the recent lockdown has brought is the more rapid digitalisation of industries that had hitherto been somewhat reluctant or resistant.
And as more and more businesses expand their digital horizons, so the adoption and application of ML will increase accordingly.
Because while take-up has initially been slow, momentum is growing. Universities have started degree courses in ML to produce the data scientists the country needs to exploit these technologies.
Amazon Web Services and Microsoft Azure both have their own ML platforms, allowing developers to create their own environments and see the power of ML in a very short space of time.
The more data businesses digitally collect, the more opportunity there is to do something clever with it – even in the most unexpected of industries.
A great example of this is, somewhat unsurprisingly, Google, which has been using AI and satellite data to prevent illegal fishing.
On any given day, 22 million data points are created that show where ships are in any of the world’s waterways. Google engineers found that, by applying ML to the data, they could identify why a vessel was at sea. This ultimately led to the creation of Global Fishing Watch, which can identify where fishing is happening and, therefore, also when it is happening illegally.
Will the machines take over?
No. While artificial intelligence and its potential for misuse is a mainstay of science fiction, in reality, ML is only as good as the data it is supplied with.
The old adage of “rubbish in rubbish out” rings even truer where ML is concerned, as it can only work and learn from the data it has been supplied with. It goes without saying, therefore, that if this data is poor, the results it predicts are likely to be damaging.
As such, the technology is very specific in its application – when it is used right, it can bring about massive benefits in increasing process efficiency.
However, its role will continue to be an assistant to improve a process rather than completely replace any human involvement.
Having said that, quantum computing does have the potential to reduce that need for human instruction, moving, as it does, away from the traditional binary state of on and off. Instead, the idea allows computers to move information around, even if it contains uncertainty – just like real life.
Along with the Internet of Things, ML is the next frontier when it comes to digitalisation. As firms begin to gather more and more data, its uses will become more keenly felt, and its true potential realised.
Ultimately, ML is about making efficiencies, learning from experience and cutting costs – and which industries can afford to miss out on that?