The text on this page was automatically translated and hence may differ from the original. No rights can be derived from this translation.
On behalf of the Ministry of Education, Culture and Science, Dialogic investigated the impact that the use of artificial intelligence can have on education in the Netherlands. The objective of the research is to gain insight into how artificial intelligence is currently being used in education (and how it will be used in the future) (What is possible?), and what legal aspects come into play in that usage (What is allowed?). The research also aims to uncover the five biggest risks and opportunities associated with this use (What do we want?).
What is possible: what are relevant applications of AI in education?
It is not easy to provide a single definition of AI, especially in terms of applications in education. Our research shows that particularly when it comes to applications that (1) automate cognitive tasks and (2) utilise large amounts of data and data-driven methods, interesting unresolved issues exist. In the educational process, teachers continuously make decisions based on their own insights regarding aspects such as the method used, the curriculum, the approach to a student, and so on. Ultimately, teachers also make some formal decisions: what is the grade, and can a child progress to the next grade? AI can assist teachers in various ways with these decisions. We identify four scenarios as most likely for the next 5-8 years: (1) AI as an educational assistant, (2) AI for learning analytics, (3) AI for personalisation of education, and (4) AI for testing.Can AI completely replace a teacher? Not in the near future.When AI becomes so advanced that it can replace the teacher, it could theoretically greatly enhance education: every student could receive 'private tutoring'. However, this stage has not been reached yet: it is expected that such artificial general intelligence (where AI matches human intelligence) will take at least several decades. Nevertheless, less intelligent AI, currently available, can support teachers in such a way that they can spend more time per student or work more efficiently.
What is allowed: what are the legal hurdles in the application of AI in education?
The application of AI in education touches upon various generic regulations. Some of these regulations are non-mandatory, meaning they can easily and often be deviated from by contract, such as the regulations of Copyright Law and Database Law. The flexibility of these private law regulations is much less present in public law regulations, such as the rules on transparency and reuse of administrative or governmental information. The applicability of these regulations can deter developers from forming partnerships with institutions if it means their knowledge will be exposed. Additional contracts on intellectual rights may partially assist in this matter. However, it must be ensured that collaboration with market players does not lead to the creation of monopolistic positions that could violate competition rules. Decisions made by institutions are naturally fully governed by the general rules of administrative law, including the general principles of good governance. Applying these principles to AI applications can be challenging, as they inherently clash with the 'black box' that AI creates. Nonetheless, this does not seem to pose an insurmountable problem. The regulations on the protection of personal data (GDPR) may indeed be a stumbling block. The nature of personal data in education and the nature of AI do not legally align well, and it seems very difficult to envision widespread AI use without legal regulations. Of course, this changes entirely if the data used is no longer personal. If issues arise, there do not seem to be any significant problems in applying liability rules to AI applications. This regime is very open and flexible so that those harmed by AI applications do not have to be left out in the open. The education-specific sectoral regulations allow for a lot of freedom and do not seem to contain any strict prohibitions. However, actions must always be taken in line with the principles of education law, with the interests of the child being a central ethical focal point, considering whether this is the case with AI. Practically speaking, sectoral regulations pose an obstacle: there is no strong top-down control over the educational content, apart from the fact that this is a very sensitive issue. Bottom-up initiatives are also difficult: while initiatives can arise in individual institutions, the collaborative model and the variety of regulations at the institutional level are likely to hinder a large-scale movement. This is disregarding the potential aversion of teachers (and institutions) whose interests and positions can directly be affected by the use of AI.What do we want: what are the key opportunities and risks of AI application in education?
Opportunities of AI applications in education
The research indicates that applying AI in education offers opportunities to achieve the following positive outcomes:- Reducing the workload of teachers through AI application for (administrative) tasks.
- Personalised learning: better tailoring education to the student, with improved outcomes and a better learning process.
- Supporting teachers with holistic, evidence-based insights (learning analytics).
- Improving the method of testing knowledge.
- Enhancing the effectiveness of digital learning tools, also in synergy with other technologies such as VR and serious games.
Risks
The application of AI in education could potentially lead to the following negative effects, unless mitigations are put in place:- Some educational goals may be compromised by AI if the focus on technology and the aspect of 'knowledge transfer' becomes too dominant.
- Bias in humans translates to data, which is then adopted by AI.
- A diminished outlook for the profession of 'teacher' (for current and future teachers).
- Dependence on black-box models (unexplainable AI) versus the responsibility of the teacher.
- Application of AI while other basic provisions are not yet in place.
- A shift in power dynamics among producers of educational materials.
Recommendations
In order to promote the use of AI applications in education, there are several aspects on which policies can be formed, and/or on which policymakers (specifically the Ministry of Education, Culture and Science) can take action:- Promoting the acceptance of AI in education by teachers, students, and parents.
- Enhancing the digital skills of teaching staff.
- Establishing a data infrastructure.
- Facilitating experiments using AI in education.
- Promoting a multidisciplinary approach to AI development.
- Developing a quality mark for responsible AI application in education.
- Initiating further research into AI in education.
