Education Futures Episode 1: Short Term Growth of AI in Educations
AI, or artificial intelligence, is a term we hear tossed about frequently these days.
In general, we use the term to talk about machines that seem to replicate human intelligence or at least the thinking functions we normally associate with humans, like learning or problem-solving.
Essentially, we use AI is a broad, catch-all term to refer to any computer system or software that is able to interpret and learn from external data and then use that “learning” to perform tasks and adapt its performance over time. Most of us use AI systems every day, whether it’s through our digital phone assistants and home managers like Siri and Alexa, support chatbots, or via more ambitious applications such as autonomous or assisted driving platforms.
Now, there are a number of different types of AI systems or approaches to creating AI applications. One such approach to AI is machine learning. Machine learning uses rules or algorithms to sift through data, learn from it, and then make predictions about something. Whereas traditional software programs consist of specific sets of instructions to accomplish predefined tasks, in machine learning we actually “train” the machine, using rules and large amounts of data, so that it can learn how to perform the task. For instance, instead of trying to explain how a cat looks like to a machine learning algorithm, we simply provide it with millions of pictures of cats. The algorithm then finds recurring patterns in those images and figures out for itself how to define the appearance of a cat. Afterward, when we show the program a new picture, it can distinguish whether it contains a cat or not.
A subset of machine learning that we often hear about is called deep learning. Deep learning is a machine-learning technique that uses multi-layered artificial neural networks to teach computers to learn by example. Deep learning is the technology used for training autonomous vehicles to recognize a stop sign or for voice recognition applications that allow us to talk to our phones and TVs.
Machine learning and deep learning require large data sets to learn effectively. and while they can process and learn from both unstructured and structured data, the more structured data there is to work with the faster the system can evolve. In other words, an AI system can process and learn to use a structured information domain like introductory biology, with its established taxonomies and vocabularies, much more efficiently than, say, a data dump of a few million seemingly unrelated financial documents.
In education, different AI approaches are currently used for everything from adaptive and personalized learning platforms to online tutoring programs. There is indeed plenty of hype, but what’s the real promise? More importantly, what kind of AI applications make the most sense in education’s short-term future?
A simple way to answer this question is to say that, in the short term, AI in education will be most successful when it is used to work with structured data and clear rule frameworks. The best applications will take advantage of well-established content domains and focus on activities where we have enough experience and data to actually “supervise” AI learning.
Given that context, here are some areas where I believe AI will have the greatest impact in education in the coming 1-2 years.
Tutoring and Support Assistants (Chatbots) — Intelligent chatbots have already proven effective at providing just-in-time subject-matter answers, engagement feedback, and general support help in online learning environments. As natural language processing capabilities continue to improve, we can also expect to see increased use of voice-based assistants that can respond to queries in multiple languages.
Content assembly and production — Modern education is dependent on foundational content or information. Beginning in the second half of the 20th century, this foundational content for most undergraduate courses came in the form of textbooks. In recent years, much of the information, particularly for general education courses, has become increasingly well-defined, been given standardized structures, and been openly licensed. This means that AI can be applied in these areas to automate the generation of core content, study guides, and quiz items. AI can also be used in media production and course assembly, complete with interactive content.
Content Curation — Given the structured information contained in course syllabi and domain taxonomies and vocabularies, it is now straightforward work for AI applications to search targeted resource libraries, as well as the general web, to curate and assemble course content.
Grading of High-Effort Learning Assignments — As online learning moves beyond linear media experiences with low effort, multiple-choice questions, AI will be employed increasingly to interpret and respond to open-ended, rubric-based, free-text assignments This will allow us to expand the use of critical thinking assignments in online learning while reducing the amount of human grading and feedback required.
Course Individualization and Adaptation — AI can be applied to online content to present course content in a sequence and at a pace that makes the most sense for individual learners. AI tools can also be used to remediate content for students who are hearing or visually impaired. This opens up possibilities for students who might not be able to attend school due to illness or who require learning at a different level or on a particular subject that isn’t available in their own school. In this way, AI can help break down silos between schools and between traditional grade levels.
Globalization and Localization of Online Content — We will continue to make inroads with natural language processing and AI-driven language translation. This will help us cross language and geographical barriers with online learning and also allow us to connect diverse populations of learners within global learning networks.
– Rob Reynolds, Ph.D.
Further Reading on AI and AI in Education: