The Procurement Jargon Buster: Explaining AI, Machine Learning, RPA, NLP, Blockchain, Big Data – Industry 4.0

15th January 2020

Article 1 in our 3-part series – An overview of Artificial intelligence

Introduction

For almost every industry one can think of, disruptive technologies such as AI, RPA, Bots etc have begun to have a noticeable impact.  In the past 10 years, we’ve seen investment in these technologies increase exponentially, with £850m being invested in UK companies alone in the first six months of 2019 (a 600% increase on the total investment in 2014).  Within the purchasing space, AI is being embraced in a number of ways, from delivering a reduction in/automating transactional activities, providing enhanced data analytics, improving assessment and management of supplier risk, and (perhaps more creatively) can be seen in the development of procurement bots as day to day purchasing assistants.

It’s difficult to attend a procurement conference these days without a reference to technology being “a game changer” for the profession, or finding high on the agenda the ever changing role of the procurement professional and how this will be influenced by technology both now and in the next 10, 15, 20+ years.  Often however, it is assumed terms such as Artificial Intelligence, Robotic Process Automation (RPA), Natural Language Processing, Blockchain (to name but a few) are fully understood by the audience when the reality is these concepts are often inaccessible and difficult to understand.  The purpose of these articles is to help shed some light on the dark art of AI and the many associated offshoots, and to provide some genuine user cases to help understand how AI works and what it could mean for you.  In subsequent articles we will explore in more detail the concepts of machine learning, RPA, NLP, Blockchain etc but for now, we’ll focus on the basics…

 

In the first 6 months of 2019, AI funding in the UK had already surpassed 2018’s figures. AI investment reached $1,021,642,595 in 2018 in the UK, but reached $1,063,012,777 in the first 6 months of 2019…
Source – UK artificial intelligence investment reaches record levels as UK AI scale ups tackle fake news and climate change

 

So why is AI such a big deal?

AI allows us to be more efficient and effective in many ways.  It can solve complex problems using large data sets that would be impossible for a person to achieve.  It can process and manage millions of data sources in real time providing meaningful outputs.  It can identify underlying patterns in data and information in a way that a person would be unable to do effectively.  It can complete tedious and laborious tasks in super quick time, often with 100% accuracy.  It doesn’t eat or sleep and can continue to work in the background 24 hours a day.  And most importantly, it can learn.  From feedback.  From instructions and direction.  And from its own mistakes.  Allowing it to improve continuously over time.

 

“AI is here today; it’s not just the future of technology. It’s embedded in the fabric of your everyday life.” —Neil Jacobstein, Singularity University Chair, AI & Robotics

 

But what is AI exactly?

AI stands for “Artificial Intelligence”, but this is quite a broad term and in fact, there are many variations on AI/AI based technologies, all of which work in different ways, achieving different outcomes.  What we typically think of when we hear the words Artificial Intelligence (or AI) is a million miles away from the reality.  Our thoughts immediately run to human-like robots (androids), with what is known as “Strong AI” or Artificial General Intelligence (AGI).  This is essentially AI that can operate at or near human level capacity in respect of problem solving, creative thought and learning, demonstrating real sentients and awareness of surroundings.  The most optimistic projections suggest this is 40+ years away from reality.  This is not the AI that influences our everyday lives so, for now at least, we can focus our attention elsewhere.

The second form of AI is known as “narrow AI” or “weak AI”.  This is what we interact with most days.  This type of AI adheres to predefined logical rules (algorithms) and behaves in a “predictable” manner.  There are two main types – Supervised AI and Unsupervised AI (with a further sub-type of semi-supervised but we’ll keep it simple).

 

“AI is perhaps the granddaddy of all exponential technologies—sure to transform the world and the human race in ways that we can barely wrap our heads around.”
Jason Silva – The Exponential Guide to Artificial Intelligence

 

Supervised AI works in the same way as teaching a child.  For a child we might give them a series of sums, explain what the terms “add”, “subtract”, “divide”, “multiply” mean and how to apply them, and check that they got the answer right at the end.  With AI it’s quite similar.  We feed in a bunch of inputs.  We provide some rules on how to interpret those inputs for each scenario.  And we check the output and make amendments until they correspond to expectations.  For example, your phone may categorise your pictures in to a “holiday album” having recognised pictures of the beach, swimwear, location tags etc. without having to be told.  But to do this, Apple/Google has taught the AI what to recognise and what to exclude. In a purchasing context, this could be processing semi-structured spend data using key words or phrases allowing spend to be grouped against pre-defined categories.  We call this “supervised AI” as we know the expected output and we correct it if it’s wrong.  (Such as if Dell Notebooks were categorised as “stationery” for example.)

Unsupervised AI works differently.  We may know what the inputs are, but there may not be an agreed or expected output by which to check the answers.  The simplest way to think of unsupervised AI is to think of it as enhanced pattern recognition.  The AI can spot underlying patterns to unstructured data which we would typically miss.  There’s no right answer; only patterns that the AI has found for us to consider which it then clusters according to common characteristics.  This is particularly useful when we don’t know what we’re looking for!  In a purchasing context, this could be used to identify underlying determinants or factors in price.  For example, a relationship between all call-off contracts agreed in March and securing a lower price may be identified as a common characteristic of these agreements (potentially driven by it being the financial year end).  And in recent times, one of the most beneficial use cases has been in supplier risk, where AI allows us to process millions of data points simultaneously (covering press articles, financial account, performance ratings, social media etc), facilitating dynamic risk scores which a human team could never achieve.

 

Still not sure? Watch the video below from Edureka

 

So, what is Machine Learning, RPA, Blockchain etc?

Hopefully this article has provided you with a better understanding of what AI is and the benefits it can bring.  In the next article, we’ll explore in more detail concepts like machine learning and natural language processing, as well as solutions developed based on an AI architecture such as RPA, Blockchain, Bitcoin etc.  In our final article, we’ll look more holistically at Industry 4.0 and the implications this has on the purchasing profession in the short, medium and long term.

In the meantime, if you would like to discuss any of this further please get in touch 

 

get in touch

Please fill in the form below.