AI is creating huge opportunities for workplace learning and employee development. In his new book “Artificial Intelligence for Learning”, member of OEB’s Global Council Donald Clark discusses how to use it to improve all aspects of learning in the workplace. We wanted to understand the difference AI can make in organisations and where it is best implemented so caught up with him.
Artificial Intelligence has been quite the buzzword/hype and is a big technological trend for some time already. Can you summarise in a few sentences what AI for learning really means?
AI is the technology of the age. It is everywhere, in every device and mediates almost everything you do online; Email, Facebook, Twitter, Instagram, Google, Google Scholar, Maps, YouTube, Facebook, Twitter, Instagram, Netflix, Amazon…. all mediated by AI. It would be bizarre if it were not to have a major role in online learning… and it does.
AI is already in online learning. Almost all informal online learning found on Google, YouTube and other sources of learning are searched for and many mediated by AI. Even in formal learning, AI is now being used in real organisations to engage learners, support learners, interface through smarter interfaces such as voice, create content, curate content, adapt content, personalise and assess learners. This is the core message in my new book “AI for Learning” where I run through these applications, across the entire learning journey, with real examples, to signpost towards this new world of online learning.
This leads to the question of how AI can best support employee development. In which areas can it be implemented?
AI is already being implemented in online learning in the workplace. Whether it is formal learning through recommended learning paths, chatbots and learning in the workflow, we are seeing a shift towards data-driven ecosystems of learning, with LMS, LXP and data-driven approaches to the management and delivery of organisational learning. I’ve been involved with this for several years and have no doubt that the data-driven approach, sometimes through a Learning Record Store like Learning Locker, can transform an organisation that delivers courses, into a true learning organisation by making it more sensitive to the real needs of learners, employees, managers and organisations using data.
I discuss this shift in the book, along with the changes in skills within Learning & Development and Interactive Design teams. We’re shifting from linear to complex, monologue to dialogue, surface to deep learning, old presentation software to smart software. This requires a different mindset and skills set.
It is also important that we understand what data is and how it can be used to describe, analyse, predict and prescribe action in learning. The worst outcome is to get stuck with dashboards. Data is about action not reading dashboards. I spent some time on this topic in the book as it is the one area that is least understood and therefore poorly implemented.
What are some of the biggest challenges when implementing AI for learning?
There are two sorts of challenges – psychological and physical.
Top of the psychological list is ‘mindset’. You have to get your head around this new world, one where much of what is happening is invisible. Most people are blissfully unaware of AI and its ubiquity. It is everywhere online, filtering spam, stopping dick-pics, porn and hate speech, compressing and decompressing files, selecting things for you, so that you are not flooded by a tsunami of information. I liken AI to the the bottom bulk of an iceberg, which now keeps the visible front of the internet afloat. In AI-driven learning you have to adapt to the idea that much of the efficacy is invisible, AI mediates the learning, helps engage, support, give feedback, scaffold, adapt and recommend throughout the learning journey. Another psychological angle, is to stop obsessing with AI and ethics or at least anthropomorphising. Sure, there are ethical issues around AI, as there is with all technology – we drive cars but they kill 1.3 million people per year and seriously maim many more. I discuss bias, transparency, gender, race, employability and even the existential threat in the book. In learning, however, the ethical issues are hugely exaggerated. Human teachers are far more biased than AI and with AI we can work to deliver learning that is truly blind to race, gender and social class. It’s fundamentally a design problem.
Physically, you need to know its limitations and delivery parameters. By this I mean be careful with tapping into Watson, Microsoft or Google AI services – you will have no control, little customisation opportunities and it will be costly. That’s why we package all of our WildFire code and functionality into SCORM packages, so that clients remain in control. You need to work with people and organisations that know what they’re doing. Funnily enough, in making things simpler on delivery, you have to know a lot about the complexity.
In your book you also give case studies and real-world examples of companies that successfully use AI to improve learning in the workplace. Can you share an example?
My favourite example is one that will surprise many people, as it is around using AI to create content. We produced a very large amount of useful, online learning for TUI, using WildFire. This was hugely successful in the organisation. Not only did we produce the content in record time (as the AI did most of the heavy lifting), we did it at 10% of the market cost. Interestingly, we had not a single face-to-face meeting and this was before Covid. When tracked through to use, we also showed proven efficacy through increased sales. I loved this project as it paved the way for agile production, proved that the technology worked, saved TUI a pile of money, was super-quick to complete and had huge impact in the organisation. I give a lot of real examples in the book, where I think AI has really advanced thinking by actually delivering efficacy.
Jobs will be automated and students need to be educated for currently non-existing jobs…. The link between AI and the future of learning & employment is complex. How do you see this?
Again, I talk about this in the book, and in truth no one knows for certain. But we have some pointers. First, before COVID-19, we had pretty healthy levels of employment, despite the spread of AI and robots in manufacturing. When radical technology comes along, the steam engine, mechanical loom, electricity, computers and so on, we humans adapt well. Technology has given us the freedom and wealth we now enjoy, it will continue to do so. How that will be equitably distributed is another question, a political question.
Some of the research has already been proven wrong. Frey & Osborne has been liberally quoted at OEB since being published in 2013 but we’re seven years into their forecasts, which were clearly wrong. Other quotes I’ve seen, such as “‘65% of children entering primary school today will be doing jobs that have yet to exist” was simply made up. AI changes job roles rather than eliminate jobs. Uber has changed the skills taxi drivers need towards running a small business, using an app and customer care. It hasn’t eliminated taxis.
There’s little discussion around AI on employment in education and training but I tackle this in the book. First, forget about robot teachers, that’s a fallacy. Online learning will, to quote a couple of movies, be more ‘Her’ than ‘R2D2’, sophisticated, scalable online services that get better with use. We see hints of this with Duolingo. It is already happening in adaptive learning.
The OEB community is open to looking “over the horizon” in order to shape the future of learning. “Where to” will AI for learning go next?
If we want online learning to improve, it must get smarter. For 30 years we’ve been largely delivering rather flat and linear media. We saw, disappointingly, during the COVID crisis. Smarter online learning needs smart software and everyone agrees that AI is smart. The Top Ten tech companies are all, essentially, AI companies. The Top Ten online learning companies will, at some point, all be AI-driven companies.
We already see this happening in China and the US. A US University liked one platform I’ve been involved with so much they took a huge stake and are now delivering entire degrees using personalised, adaptive learning. Companies like Learning Pool have acquired an AI company and invested in AI to deliver the next generation of workplace learning tools. They have seen huge, recent growth.
I finish my book with some speculative thoughts around non-invasive and invasive tech… brain tech. Early days, but what we’re seeing in helping the disabled with implants is already being seen as pointing towards learning and brain prostheses. In truth, AI will simply creep into everything in learning technology. In many cases, it already has. Its invisibility is its strength. It quietly produces better interfaces, delivers the right thing to the right people at the right time, supports and scaffolds learning, can create and curate content, even assess. It’s time is here.
Donald Clark, member of OEB’s Global Council, will be speaking at Online Educa Berlin 2020. “AI for Learning” is available on Amazon and other online book sites.