MOOCs have gone from being hailed as the future of higher education - a revolutionary tool that would make traditional institutions obsolete - to being written off as an unsustainable fad with enormous drop-out rates. In this short presentation, I share a few thoughts on how AI can support MOOCs in reaching their full potential as a tool for democratising education.
1. Can AI save MOOCs?
How personalised machine learning can
improve the way we learn online.
2. The history of the MOOC
The term MOOC, for ‘Massive Open Online
Course’, was coined by Dave Cormier and
Bryan Alexander in 2008 after the launch of a
free online based course from Athabasca
University in collaboration with the National
Research Council.
3. The history of the MOOC
In 2011 Stanford University took the MOOC to a
much larger scale with the launch of a course in
AI that saw 160,000 students enrolled.
While some were sceptical and dismissed
MOOCs as a fad, others hailed it as the biggest
revolution in education since the printing press.
4. The history of the MOOC
In the years that followed, a plethora of MOOC
providers established themselves, including
Coursera (Stanford), Udacity, MITx, edX
(Harvard) and Future Learn.
Yet, the predicted revolution failed to arrive.
So, what went wrong?
5. Research suggests that there are still too many issues
with MOOCs for them to have the disruptive impact on
education systems that some thought they would
have. These issues include:
➔ Lack of community
➔ Limited personal learning experiences
➔ Lack of engagement and follow-through
➔ Issues around timeliness in course delivery
➔ Failing teacher support
6. 1. Lack of community
With 100,000+ students enrolled in
some courses, MOOCs have struggled
to create spaces that feel intimate and
that promotes the creation of
communities of practice (CoP).
Posting in large, unmonitored
discussion forums can feel more like
shouting into a void than engaging in
constructive conversations.
7. 2. Limited personal learning
experiences
The one-size-fits-all delivery of
content and assessment options can
give MOOCs the feeling of being mass
produced with learners struggling to
feel seen and have their individual
needs acknowledged.
In its current forms, MOOCs also have
limited ability to take into account
students’ preferred learning styles and
possible barriers to access, such as
disabilities.
8. 3. Lack of engagement and
follow through
The free nature of many MOOCs, and
the ability to enrol in limitless classes,
has seen many people sign up for
courses that they fail to be active
participants of.
These non-participating students even
have a name; lurkers. While they might
log in and read other people’s posts,
they don’t comment and fail to
participate in discussions.
9. 4. Issues around timeliness in
course delivery
Many MOOCs are self-paced, which
can be appealing for adult learners
wanting autonomy and control over
their studies.
However, the ability to enrol in any
course at any time can be
overwhelming and result in a lack of
structure and motivation to finish.
10. 5. Failing teacher support
The aforementioned factors of
massive student cohorts and self-
paced learning has made it difficult for
instructors to offer individual support in
a timely manner.
Peer support can also be lacking as
discussion forums are often large,
unstructured and with limited
possibilities for searching and sorting
results.
11. This has resulted in massive
dropout rates.
(Data suggest that up to 90% of students
enrolled in MOOCs don’t finish their course) Note
These numbers are
inherently hard to pin
down, as people might
complete the course,
but choose not to pay
for a certificate - as in
the system of Future
Learn.
12. Can AI help turn
this trend and
see MOOCs
realise their full
potential?
A note on
terminology
AI is used as an
umbrella term that
includes concepts like
machine learning and
virtual learning assistant
technology.
13. Lack of community
Upon enrollment, AI could see institution gather basic demographic data
like age, nationality, profession, education levels, barriers to accessing
information (like physical or cognitive disabilities) as well as learners’
motivation for enrolling in the MOOC.
In addition, students can be asked to take a test that can help further
personalise their user profile. These profiles can then be used to
automatically sort learners into groups of like minded people with similar
interests and preferred ways of learning.
Creating supportive
learning communities
14. Limited personal
learning experiences
Following the creation of learner groups, course materials can be created
and presented in ways that suit each group to optimise accessibility and
engagement.
By the end of the course learners have accessed the same information,
but the materials have been delivered in bespoke to each learning group.
Throughout the course learners also have the opportunity to interact with
peers within their learning group, where the chances for meaningful
discussions and peer support are increased.
Personalising learning
pathways
15. Lack of engagement
and follow through
By using AI to register and track learner engagement on the platform,
including how often learners log in, which modules they have completed,
what learning materials they have downloaded, and how often they post
or interact with others on discussion boards, MOOCs can send timely
prompts to learners that help boost engagement.
By producing data on interaction with individual modules and learning
materials, AI can also help alert course providers and instructors to areas
where drop-out rates spike.
Tracking and optimising
engagement
16. Issues around
timeliness in course delivery
With thousands of free courses from multiple providers, long enrolment
periods and flexible, self-paced schedules, it can be hard for learners to
find courses that offer “just in time” training.
By using AI to scan learners’ professional profiles on LinkedIn and other
platforms, course providers can tailor marketing of courses based on
activities like changes in professional roles, shared/liked content or group
memberships.
As such, this supports adult learning theory, which suggests that adults
are more likely to engage in learning when they can see immediate use
for their new skills.
Timely course offerings
17. Failing teacher support
Huge student cohorts, flexible enrolment dates and self-paced learning
makes it near impossible for instructors to be on hand to support learners.
AI could address this through the introduction of smart chat-bots that can
read and answer student questions in a timely manner.
For questions that cannot be answered by the bot, the instructor could be
automatically notified. Further, these bots can read discussion forums
and, utilising the power of the learning groups, prompt students to help
each other.
Enabling ongoing teacher
and peer support
18. Other roles for AIs can
include
Personal tutors who give learners quizzes and monitor
questions asked in forums to determine their
performance and map areas of strength and
weakness. This data can then be used to supply
learners with tailored materials that support them and
place them in Vygotsky's zone of proximal
development.
Career development advisors who send automatic
suggestions for which courses to enrol in next to build
on their previous learnings and help them progress in
their career or field of interest
19. But there are
downsides
There is a strong correlation between data gathering
and the effectiveness of machine learning. In the case
of MOOC students, that means that the more data that
is gathered and stored, the more tailored the learning
experience and support can become.
While some data is standard administration info,
others, like access to healthcare records to create
bespoke solutions for people with disabilities, or even
cross reference activities on students social media
accounts, come with considerable privacy and data
security concerns.
20. It is clear that to reach their full potential as tools
for free and accessible education on a global
scale, MOOCs must become more intimate,
engaging and supportive.
Whether AI can help make that a reality
is not a question of technical ability,
but of our willingness to sacrifice
privacy and data security to gain
improved learning experiences
online.
The question, then, is perhaps
not can AI save MOOCs, but
should it?
21. Texts and videos that inspired this presentation
Alexander, B. (2019), 5 AIs in search of a campus. Retrieved from
https://er.educause.edu/articles/2019/10/5-ais-in-search-of-a-campus
Department for Education (2014). MOOCs: Opportunities for their use in compulsory-age education.
Retrieved from
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/315591/DfE_RR355_-
_Opportunities_for_MOOCs_in_schools_FINAL.pdf
Kellermann. D. [Microsoft]. (2019, July 18). Demo: Teams in the Classroom at Microsoft
Inspire 2019 [YouTube]. Retrieved from https://www.youtube.com/watch?time_continue=34&v=NcbQ2UK69Tc&feature=emb_logo
Knowles, M. S., Holton, E. F., & Swanson, R.A. (2015). The adult learner: the definitive classic in adult
education and human resource development [ProQuest Ebook Central] (8th ed.). Retrieved from
https://ebookcentral.proquest.com/lib/qut/detail.action?docID=1883897
Laaser, W. (2015) “The rise and fall of the ‘Massively Open Online Courses’”, South Eastern European
Journal of Public Health (SEEJPH). doi: 10.4119/seejph-1804 .
O'Donnell, E., Lawless, S., Sharp, M. & Wade, V. (2015). A Review of Personalised E-Learning: Towards Supporting Learner Diversity.
International Journal of Distance Education Technologies, pp. 22-47. Retrieved from
https://www.scss.tcd.ie/seamus.lawless/papers/IJDET-2015.pdf ‘
Rivard, R. (2013) Measuring the MOOC Dropout Rate. Retrieved from
https://www.insidehighered.com/news/2013/03/08/researchers-explore-who-taking-moocs-and-why-so-many-drop-out