Artificial intelligence, or the rise of the smart machine, is the stuff of science fiction. But applying machine learning to apprenticeships can deliver a better learning experience, argues Mark Abrahams.

 

The late, and great, Stephen Hawking once told the BBC that “The development of full artificial intelligence could spell the end of the human race…Humans, who are limited by slow biological evolution, couldn't compete, and would be superseded."

Dystopian fears are widespread when it comes to artificial intelligence, from complaints about excessive monitoring in the workplace to robots being all too human.

Increasingly, however, we are coming to understand that smart machines – machines that can replicate some human behaviours such as learning and problem-solving - can help human societies improve what they do.

So what does this mean for apprenticeships?

Running an apprenticeship programme involves a lot of administration – time that could be put to better use teaching apprentices, whether face to face or online. Resources are scarce. And the Skills Commission noted that around 30% of people who start apprenticeships do not complete them.

As Jisc’s CEO Paul Feldman commented recently, we need better data if we want to make the apprenticeship system work.

Can artificial intelligence help workplaces deliver these schemes more efficiently?

 

How we use artificial intelligence

Organisations across many business sectors are now using machine learning – meaning the current scientific methods we use to build artificial intelligence - to improve efficiency and their understanding of their clientele.

You may think that AI is distant from your everyday life, but in fact, AI affects our travel, workplace and shopping choices. And the reach of AI is growing. Statista estimates global revenues from AI market worldwide will grow from $3 billion in 2016 to nearly $90 billion by 2025. Global revenue from ‘big data’ was $122 billion US dollars in 2017 and is predicted to grow to $200 billion by 2020.

Computing technology has advanced so rapidly over the last few years that data analysis that was previously time-consuming or needed expert input can now be generated automatically. Powerful new techniques can now make predictions more consistently and accurately than humans.

Many readers will be familiar with virtual assistants such as Alexa or Siri, which can readily listen to voice instructions and act on them. These applications rely on big data techniques which convert speech to text, which is then analysed, ‘understood’ and responded to.

Another less known use is voice analysis in call centres to highlight the emotional state of customers on the phone – presenting this information back to customer service representatives improves the effectiveness of customer interaction.

Within the recruitment industry, algorithms have been developed which can interpret and screen applicants’ CVs. Similar methods are being used to score short exam and test answers – for example, the SAT written test in the USA.

Some businesses use machine learning to automatically score video answers provided by job applicants to measure their job suitability.

These are all applications of readily available machine learning techniques which can process and make sense of unstructured data, for example, speech, video, images and free text. The features of the data are compared to existing data and used to make robust predictions.

 

Predictive analytics and apprenticeships

Predictive analytics (PA) is about helping us make projections about future events through uncovering patterns and relationships (though it cannot necessarily be 100% accurate).

It sounds like traditional statistics, but it isn’t. Computer algorithms can utilise hundreds, even thousands, of data points to find the particular data patterns which predict an outcome.

Unlike traditional statistics, predictive analytics identifies the various multiple profiles or combinations of factors which might lead to, for example, the likelihood that a learner will drop out.

For apprenticeship schemes, training providers and employers need to be able to measure trends in learner engagement and motivation to prevent attrition. We can solve this problem with predictive analytics. PA can also provide high-level reporting of retention risks across groups of apprentices.

PA can also be used to support tutors and other team members, such as managing workloads. It can look at patterns in the responsiveness of tutors, how quickly work is assessed and returned to learners and whether particular criteria are difficult to adequately evidence.

For larger schemes, PA could potentially be used to match tutors to learners.
Our team of data researchers and psychologists at MWS Technology Ltd. have developed Aptem, a one-stop apprenticeship management app with built-in machine learning functionality.

Aptem can be used to identify high-risk programmes or learners through its Early Warning System. It uses Predictive QAR to allow users to see how their QAR scores might develop and intervene in time.

And because it is an ‘end to end’ system, meaning that it manages the apprenticeship programme from start to finish, it generates rich data which can be used to assess the pace and progress of personalised learning and make this information readily available to Ofsted.

 

Forward to the future

Ginni Rometty, chair, president, and CEO of IBM, said of AI that “this technology will enhance us. So instead of artificial intelligence, I think we'll augment our intelligence.”

I agree with the idea that AI evolves our society. And I’d argue that using AI in apprenticeship learning shows how AI can enhance the delivery of the programme and the learning experience, to make sure we deliver the highly-skilled workforce the world needs.

 

 

Mark Abrahams is Head of Research at MWS Technology Ltd. He is a Chartered Psychologist specialising in talent assessment & development, psychometrics and analytics.

 

         

Top of Page