Artificial intelligence in education
Reading time ca. 7 minutes - Read, like, comment!
This article appeared first in the Journal "Bildung für Europa" of the NA beim BIBB.
This article was originally published in German by Markus Palmén.
Artificial intelligence promises to be a great democratizer of education, bringing customized learning paths to masses. Adult educators need to understand the technological and ethical dimensions of the megatrend to fully participate in co-creation of quality educational AI.
Artificial intelligence and adult education. Pause, for a moment, to think of the mental pictures these words conjure up. You may be imagining a futuristic scene of computer-assisted learning or thinking of current everyday life, like the AI-powered smartphone in your hand. Whichever the association, artificial intelligence (AI) is a megatrend in education.
Artificial intelligence mimics human learning
As a term, AI refers to computers capable of mimicking human learning and autonomous problem-solving based on those learnings. The technological applications of AI range from powering internet search engines and online news and advertising algorithms to driverless cars.
In an education context, applications are equally abundant. A voice assistant on a smartphone performs information searches for its user. An educator leaves the grading of a multiple-choice test to an AI programme. An online AI-powered study interface follows a learner´s progress and suggests bespoke assignments based on the learner´s challenges. More advanced yet, emotion detection software records facial expressions of a student and starts to recognize where the learner struggles most.
AI promises much in education
From the examples above it becomes clear that AI operates on two levels in learning: it is a learning aid used by the learner but also an observer of the learner, operating behind the scenes on the meta-level of the learning process. We can attempt to summarize the “roles” and “promises” of AI in learning as follows:
- AI as a “teacher” and learning assistant promises to be a great equalizer of education, bringing cost-effective education to the masses and enabling mass distance education.
- AI as an educator´s technical assistant in mechanical duties such as grading and back office tasks promises to free up time for the educator for personal interaction with students.
- AI as an observer of learning, pinpointing individual learners´ progress and learning bottlenecks, promises to customize and individualize learning, offering a wealth of learner data to educators to help them improve their didactics.
The last “promise” on the list - customizing- is especially significant for adult learners with heterogenous prior learning backgrounds and needs.
Paradoxes and privacy concerns
However, lists like the one above run the risk of simplifying what is at heart a complex issue with strong ethical undercurrents. Upon closer analysis, there is a curious paradox at the heart of the debate about AI in education. AI holds the promise of democratizing lifelong learning, of bringing personalized learning to the masses. At the same time, however, lifelong learning is seen as a strategy against an AI-dominated future where intelligent machines make a large chunk of human jobs obsolete. In other words, from a policy perspective, lifelong learning is a strategy of “future-proofing” the workforce against AI. Being both a poison and a medicine is an irreconcilable paradox, and one that simply needs to be acknowledged as testament to the complexity of the phenomenon.
Furthermore, focusing mostly on the pragmatic aspects of AI risks bypassing the ethical considerations inherent to it, which are a dominant part of the current AI debate. AI algorithms learn by analyzing large amounts of data. For example, educational AI feeds on huge amounts of observations of different learners, individuals. The questions at the heart of the ethics debate therefore are: who is actually collecting all this data? Who owns it and has the right to access it? Is the data of learners protected somehow or can it even be harmful to them? Imagine, for example, that data collected during a company training creates a “class-system” of fast learners and underachievers. This debate echoes that surrounding the Big Data companies such as Facebook, Google and Amazon, also relying on AI algorithms in their business operations.
Concerning this privacy debate there are promising political developments, at least in Europe. In November 2019 EU education ministers concluded that AI use in lifelong learning must be “ethical” and “human-centric”, granting people access and a degree of control over their own AI-gathered data. An unsurprising and unrisky political statement, perhaps, but a welcome one.
Educators as AI co-creators
Where does the adult educator fit in all this, then? I would envisage two roles for him or her. Firstly, an understanding of AI in the scope described in this article will be a staple of adult basic skills and digital literacy. Educators need to have insight of the phenomenon themselves to pass it on. Secondly, AI-education scholars, such as Professor Rose Luckin of University College London argue that educational applications of AI need to be co-created with educators to ensure their pedagogical quality.
For Luckin, co-creation does not mean that educators need to master the intricate technological details of AI. Rather, educators should feel confident in questioning AI companies who must communicate the details and data usage of their product clearly. Professor Luckin sees encouraging signs of cross-sector cooperation around educational AI. Co-creation is not widespread yet but there is an increasing appetite among educational technology companies to work together with educators. The impetus is in the right direction, Rose Luckin says.
AI – fundamentally social?
Discussing the role of educators in AI-powered education suggests the broader and often-heard question of how will AI affect the social dimension of learning.
An important distinction is helpful here. AI-assisted learning is not distance learning by default, which it may often perceived to be. AI lends itself well to distance learning applications but “personalized learning through AI” may have very social implications. Instead of online learning materials, an AI may recommend a suitable real-life tutor for you, based on your individual needs and the tutor´s skills profile. Also, if you are distance learning, an algorithm may suggest the best synergistic online learning group for you, optimizing the benefits of social learning.
As Jaron Lanier and Glen Weyl write in a recent opinion piece in WIRED magazine, the fundamental nature of AI is not independence from humans, but rather dependence on and interaction with human input. Machine learning is possible only through data of human activity. AI complements human agency, not replaces it.
Of course, the same is true of learning, a social activity by definition. Rose Luckin points out that a balance between AI-powered technology and human tuition and interaction is fundamental for quality learning in the future. A balanced mix maximizes the potential of AI and delivers on its huge educational promise, in cooperation with flesh-and-blood educators.
Über den Autor