Is the AI educator (AI-ed) coming soon? Just last week a Google engineer boastfully announced that one of their chatbots had achieved sentience (a claim that was quickly walked back by Google!), which has made some in the academy question whether we may soon be out of a job! And whilst full AI intelligence might still be some way away, the truth is that the 21st century has been home to some major improvements in AI.

In some ways, AI educators may seem to be a creation of science fiction, with portrayals from The Jetsons to Wall-E.  However, given the trajectory of developments in AI, and the fact that pre-built online educational packages have and are being used currently, AI-ed may soon be become a reality. And whilst they may not replace the professors quite yet, make no mistake, AI-ed is coming.

The stage is set for AI-ed

The AI educator is usually implemented in the realm of on-line learning because of the ease of implementation and the relatively lower costs. Although on-line learning has been a global trend since 1990, the advent of the pandemic  forced everyone on-line. Even as pandemic restrictions gradually lift, on-line learning (synchronous and asynchronous) remains popular among students.

With this in mind, in a bid to streamline processes to match the pace of change, many educational institutions have put a focus on learning/instructional design. These specialised teams focus narrowly on their area of responsibility and are able to produce an educational product (a pre-built package consisting of subject material content, enduring assessments and learning support resources) that can be implemented and taught at scale. This approach has been adopted widely in many online courses, but most notably in asynchronous on-line deliveries such as Massive Open Online Courses and on-line educational institutions and providers (such as The Open Polytechnic NZ and Open Universities Australia).

From learning package to educational intelligence

Once a learning package is adopted, this provides an underlying database on which an Artificial Intelligence can start to work. This then allows the AI to work in three main ways.

Firstly, it can perform adaptive teaching. Adaptive teaching is when AI can change its pre-determined route of delivery according to the data input and react accordingly.

Secondly, it can personalise delivery of material. Personalised delivery is when AI tweaks the delivery of the educational content according to the student’s behaviour.

Finally, and perhaps most importantly, the system can provide instant feedback is when the AI responds immediately to the student query in a relevant manner.

Furthermore, other major developments in other aspects, such as in assessments, invigilation and supporting social and peer interactions are ongoing. For instance, AI applications such as Pearson’s Intelligent Essay Assessor (IEA) have claimed to grade assessments as accurately as human experts. Even traditional educational institutions (such as Singapore’s Ngee Ann Polytechnic) are trialling the use of AI for assessment,  invigilation and content creation.

The ethics and biases of being taught by a robot

Despite the developments and advancements of AI, concerns linger.

Firstly, there are some ethical concerns with how an AI-ed might work. For instance, ethical concerns regarding ownership (e.g. who owns the data), consent (e.g. are students capable of giving genuine informed consent), privacy (e.g. level of intrusion of the AI) and biases remain. These concerns are especially relevant for AI-ed, which tend to conduct ‘dataveillance’, a practice of over-collecting data to support the operating of the AI application. Given that data protection laws are not universally comprehensive, student personal data may be subject to misuse and compromise.

This can also be a concern as biases (conscious or unconscious) of the creators may be inherited by the AI-ed. These biases may arise from the data set used to train the AI, or inability to differentiate context. And because AI can automate multiple tasks at speed at scale, inherent biases may also be automated with varying degrees of negative consequences.

Secondly, concerns about the learning experience has been well noted, especially with the level of data intrusion and de-humanising aspects that is often associated with learning through AI. Some AI applications are very restrictive in nature, forcing students into silos with minimal human interaction, and only allowed to learn along a pre-determined pathway of knowledge.

Understandably, such de-humanising actions will not improve the learning experience of any student. Summit learning, an AI educational application developed by Facebook engineers, had students walking out of the classroom in protest over the poor experience, which included the restriction of having to sit in front the computer for extended periods of time, although this is disputed by the funders, the Chan-Zuckerberg Initiative (CZI).

Finally, not surprisingly, academics and teachers worry about what this might mean for their job! Will the human teacher be extinct? Since AI-ed can create and deliver subject matter content, assess students’ learning and even facilitate social interactions, what role does a human teacher play? Many teachers can breathe a sigh of relief that the AI-ed will not be able to replace the human teacher, but the role of the teacher is likely to shift from the “sage on the stage” to a “guide on the side.”

Robot Rhetorician vs Human Professor

For now at least, beyond the development of non-academic skills (such as emotional intelligence, creativity, and communication skills), human teachers inspire and encourage using their own experiences and knowledge, which ultimately leads to a better overall educational experience for students.

Yet as the teacher shortage continues, coupled with the increased student enrolments, more educational institutions (both online and traditional) will adopt pre-built educational packages. While the AI educator is waiting patiently to take centre stage, human teachers will have to find some way to address key concerns about the AI-ed.

Because, make no mistake, once AI-ed becomes widely adopted, it may be too late to put the genie back in the bottle! So as an academy, we probably need to start thinking about this now.

Josiah Koh (The Open Polytechnic NZ), Michael Cowling (Central Queensland University), Meena Jha (Central Queensland University), Kwong Nui Sim (Auckland University of Technology)


to get daily updates on what's happening in the world of Australian Higher Education