Artificial intelligence (AI) and Machine Learning (ML) are two massive buzzwords (buzzphrases?) of the past decade (or even longer, for AI). They get tossed about with a casual flick of the wrist to indicate how comfortable and familiar we are with these terms, or with a slight sneer to indicate how passé and noughties they’ve become.
So, what’s the case? Is Machine Learning simply re-branding, a marketing polish-up of a tired and stale Artificial Intelligence? Or, beyond the cynicism, is it actually of vital use to the procurement industry? Come to think of it, what do these terms mean, and is there a difference?
And will it replace us?
Before Artifical Intelligence, there was … us
Let’s first take a moment to brush over human intelligence. We aren’t born with knowledge; Beethoven didn’t compose Symphony No. 9 still covered in vernix. We acquire knowledge because we have the ability to learn, either through experience or direct input. We are taught how to think about information, and how to acquire new information for ourselves.
Crucially, we have the innate ability to learn new skills. We can learn to dig ditches, dance, code, sing, write, fly an airplane. A human who flips burgers at a fast food chain can learn and become the chief procurement officer of the franchise and retire at 65 as CEO.
The human level of learning is not yet possible in computation. The key to understanding the difference between AI and ML is in how they learn.
ML algorithms acquire knowledge through experience. It’s the equivalent of a human working at the fast food chain for a month and realising that every Wednesday the early dinner hours were going to be packed affairs as office workers ate out to cope with hump day.
These algorithms really on big data sets to find common patterns, and offer smart solutions to very defined and specific tasks. Based on the data they’re fed, they can make decisions, statements and even predictions. A feedback loop enables learning; by being told whether its decisions are correct or otherwise, an ML algorithm can modify its approach.
On the other hand, AI acquires knowledge and learns how to apply it. ML forms part of an AI algorithm, but AI is all about taking the patterns uncovered through ML and applying the information to new and different settings.
Back to the flesh-and-bones worker at the fast food chain, it’s equivalent to them taking their experience from working at the restaurant and applying it to procurement consultancy.
AI is basically trying to mimic human intelligence, and the holy grail is known as ‘the singularity’. It’s all very science fiction, but it basically means AI will eventually be able to develop and grow like a human child. We’re nowhere near this stage yet. It’s important to not that, when it comes to procurement, the vast majority of AI is ML.
What is the role of Artificial Intelligence and Machine Learning in procurement?
Most procurement software vendors don’t make the distinction between AI and ML when marketing their products, which is why there is a misconception out there that Machine Learning is a re-branding gimmick of Artificial Intelligence.
Just to demonstrate how important many people from divergent industries think AI and ML is, here’s some stats: last year, financial-services firms invested an estimated $5.6 billion U.S. on AI, only to be outdone by the retail industry, which spend a whisker under $6 billion. And the U.S. Defence Department has plans to up its investment from $50 million to $249 million.
- Spend Classification: ML algorithms are very good at classifying things, so it makes sense that one of their best uses in procurement is sorting spend into categories and sub-categories. ML algorithms can review millions of invoices like that (cue snapping of fingers) and categorise spend automatically. As a result, ML opens procurement analysis to areas previously considered too hard with traditional methods (i.e., human power).
- Supplier Risk Management: AI can be used to monitor and identify potential risk positions across the supply chain. The power of AI mean that millions of different data sources can be screened and analysed to provide alerts.
- Anomaly Detection: ML can be used to detect abnormal behaviour in real-time data, allowing spend analysis to move beyond historical data to the here-and-now. Imagine how useful this kind of insight is during a global pandemic!
These examples demonstrate how useful and transformative AI and ML are within the procurement industry. Of course, this realisation inevitably leads to the dreaded question …
Am I going to be replaced by a robot?
If we ever achieve the singularity, then we’re all out of work. A machine capable of everything a human is that never tires and is devoid of bias? Sounds like a better alternative no matter what role you’re talking about. But at least we’ll be able to finally binge that list of Netflix shows and work on our bed sores.
In reality, though, that is a long, long way off, if it even eventuates. At this stage, it’s best to think of Artificial Intelligence as Assisted Intelligence. It’s here to help with mundane but necessary tasks, and do them a hell of a lot faster.
Buying and selling will always be a social function driven by humans; machines can’t build trusted, long-term relationships. And, besides, when it comes to learning and adapting, there’s nothing out there that beats what’s knocking around inside your skull.