AI in higher education – are we trying to fix a leaking faucet on the Titanic?

The use of powerful AI tools can provide significant opportunities for higher education. However, the challenges, such as assessment and grading, are correspondingly large. Increased use of AI has a number of challenges and raises many questions within higher education.

• Can we utilize these powerful tools for better education?
• How do we handle data protection, copyright issues and fact checking?
• What impact does the use of AI have on assessment in higher education?

The field is in rapid development, and both the higher education institutions and government agencies need to get involved and consider how this will affect digital tools for teaching, learning and assessment.

AI in Learning Management Systems (LMS)

The market for learning management systems is highly consolidated with four main players worldwide, who all seem to approach AI differently:

• Instructure’s approach with Canvas LMS is to leverage partnerships with third parties for plugins, such as the Khanmigo mentor developed by Khan Academy.
• D2L, who has Brightspace LMS, is looking at using AI particularly in the assessment area.
• Anthology has gone «all in» with AI in Blackboard Learn Ultra, introducing an AI Design Assistant which builds course modules from scratch.
• Moodle seems to be waiting to see what happens.

In the acquisition of new LMS systems, the AI factor will need to be considered.

AI in assessment

The national student survey for Norway (Studiebarometeret) for 2023 shows that 61 % of students already use AI daily. How can we ensure that this use actually enhances the students’ knowledge and learning, and is not used to shortcut and cheat the assessment process?

Although little has changed in Norwegian higher education with regards to assessment since the introduction of ChatGPT in 2022, there is an ongoing debate. There seems to be consensus that something has to change, but the solutions offered differ. The vendors of assessment tools are working on AI detectors, but there are already “humanizer” tools available to cheat these detectors. Some consider moving back to traditional on-campus lockdown exams, but these are expensive. Most agree that we need to rethink assessment, and that more continuous assessment during the year and a combination of assessment forms, is a viable path, but also expensive.

Sikt's AI tools

Sikt has developed its own AI chatbot interface, built on Microsoft Azure, with servers in Sweden. The solution emphasizes privacy, and is constructed in such a way that Microsoft has no access to information about the users. The only thing logged is the use of tokens. There are several such chatbot interfaces built, but Sikt KI-chat is the biggest consumer of tokens in the public sector.

The service is available to all customers, which includes not only higher education, but all Norwegian municipalities. Sikt is also building an AI learning hub to facilitate discussion and the sharing of experiences and practices.

Sikt is also trying out ways to create an «AI student assistant», a learning tool where bots, or agents, can be trained on specific course content and made available to students. In the early stages this will be targeted at primary and secondary education, but the plan is to expand also to higher education.

The future of AI in higher education

Right now, higher education are dealing with a lot of questions regarding AI, trying to get insight and figure out what to do.

Some of the immediate questions are:

• What are the use cases for AI in higher education?
• What are the technical solutions available?
• How do we regulate the use of AI, on an international, national and institutional level?
• How do we ensure the integrity of assessment?
• How do we build the necessary digital competencies of both staff and students?
• What are the long-term consequences of embracing this technology?

Looking further ahead there is undoubtedly a huge potential, as well as huge challenges in utilizing AI in higher education. With the current rate of technological innovation, the power of generative AI is leaping forward with every release. Tech giants are investing billions of USD in data centers and AI researchers are now talking in earnest about artificial general intelligence (AGI) possibly being just a few years away.

There’s no doubt that higher education must gain insight and competence, to be able to debate the impact of AI, and try to find a way forward that will maximize the potential and minimize the risk. It can be said that we have not yet found effective ways to meet these challenges - it's like trying to fix a leaking tap on the Titanic, without seeing the iceberg approaching.

Read more about Sikt AI-chat

The article is based on a presentation at the NORDUnet conference 2024, authored by Christian Moen Fjære and Haakon Aasprong from Sikt.