Artificial intelligence (AI) already has an immense (and often hidden) influence on what we read, watch, and buy. AI continually nudges our behavior and shapes our daily lives through content recommendations—movies through Netflix, videos through YouTube, products through Amazon—based on an AI-driven prediction of what we’re likely to want next. This core concept applies to the learning and development industry too—getting people content that’s right and relevant, more efficiently.
In this post, we’ll outline how AI is shaping the landscape for personalized learning, and steps HR professionals can take to get started and get ahead of this exponentially increasing adoption curve.
Personalization Is the Most-Wanted Learning Feature for Nearly Everyone!
Personalization as a way to establish learning relevance has an intuitive appeal, but who actually values it most? In the Global Leadership Forecast 2018 published by DDI, The Conference Board, and EY, more than 25,000 leaders picked their most wanted of over 19 different features of learning ranging from self-study to mobile-based to long-term developmental assignments. Above all other choices, personalization was the number one feature for virtually all learners: frontline or senior-level leaders; Millennial, Gen X, or Baby Boomer; based in China, Germany, the United States, Brazil, or India; in their first six months or after more than 15 years as a leader; or leading within the sales, finance, IT, or HR functions. In fact, there wasn’t a single major group of leader-learners who didn’t have it at the top of their list.It’s clear that personalization is a true unifying thirst across all leader segments—which in turn applies huge pressures on L&D managers to find new ways to meet these expectations efficiently across the enterprise. This user and business need—learning relevance via scalable personalization—is a critical role AI can play. We’ll next turn to brief descriptions of the three main types of AI-based recommendation systems applied to the context of “what’s next” for learning.
Three Types of AI-Driven Recommendation Systems for Learning Content
Popularity-Based: For these recommendation systems, the most basic type, neither the characteristics of the learning content nor the characteristics of the learner matter. AI simply serves up the most-selected learning asset from the cumulative choices of all previous learners.Content-Based: For these recommendation systems, the learning content matters but user characteristics do not. The AI system picks the next learning asset based on similarity to content a user has already picked. This similarity can be based on modality (such as virtual versus in-person), skill coverage, or many other classification factors.
Collaborative Filtering: For these recommendation systems, the user matters but the content doesn’t. The AI system picks the next learning asset based on what past learners with similar characteristics picked. These characteristics can include learner position (for example, individual contributor, frontline leader, senior leader), function (such as sales, operations, finance), and tenure, among other user profiles.
Once you’ve decided which of these approaches to use initially and longer-term, you can work backward with this end goal in mind. We’ll next turn to the key steps learning and development professionals can take to lay the groundwork for setting up these systems using AI technologies.
Getting Started Using AI for Learning: 3 Action Steps for HR
1. Classify your content: In practice, many companies start with a content-based approach to AI-driven recommendation systems. This requires curating content based on the skills it covers, the modality used to deliver it (such as web, self-study, classroom), the learner level it targets, and other key dimensions. If you use content from external vendors, ask them to align their content to your framework. Once this is in place, you can begin categorizing content into different “buckets” based on shared characteristics, which can then be used to guide AI-driven matching of learners to content.2. Know your learners: If you’re planning to use any type of collaborative filtering approach to AI, you’ll want to understand your users at the deepest level possible. For leader-learners, this can include function, level, tenure, past learning experiences, and similar factors to guide AI-based recommendation on what others similar to the current learner tend to seek out. When gathering this information, it’s important to consider diverse learner views to make sure that you’re not reinforcing past patterns of learning that don’t reflect your current state and goals for a diverse learner base.
3. (Over)communicate and build transparency: Use of AI systems can be a powerful approach for creating learning paths that meet strong user wants and rising user expectations for relevant, personalized content. However, AI systems seen as an unexplained “black box” quickly raise concerns about data use, privacy, and security. Communicate extensively well in advance of and throughout implementation of any new data-driven system to avoid perceptions of unfairness and skepticism about what information is being gathered for what purpose.
AI is shaping our lives in innumerable ways and is rapidly being folded into the ways in which leaders and employees are hired, developed, and retained. In the learning context, relevance through personalization is often the right place to start to maximize user experiences and business value from your curated learning assets. Learning professionals can act now to pave the way for data-driven approaches that meet the needs of the modern learner while also drawing on the latest AI technologies.
For a deeper dive into this topic, join me at ATD TechKnowledge 2019 for the session, Fascination and Fear: AI at the Frontier of Digital Learning. We will explore where AI can enhance your talent development efforts and where AI has limitations, and develop a plan to strategically integrate AI into your leadership development practices.
