Examples of successful implementation of artificial intelligence tools and systems in education – organisational aspect

2.6. Timing and planning

Successful implementation of AI in the education system requires a strategic, realistic, and phased approach. Rather than rapidly introducing tools into the classroom, it is essential to ensure gradual testing and evaluation of each implementation phase, with clearly defined goals and measurable success indicators. This approach reduces the risk of technical and pedagogical challenges and allows the system to be adapted to the actual needs of all educational stakeholders.

Institutions that follow this model typically develop:

• Realistic timelines that allow for thorough preparation and user training.
• Phased introduction (project approach), where the technology is first applied in a limited environment and then expanded to the entire system.
• Pilot projects whose results are used to evaluate effectiveness and adjust the strategy before broader implementation.

Example: 

In its online undergraduate and graduate programs, the University of London Worldwide conducted a pilot project integrating the generative chatbot “Walter” into law courses.

The study measured students’ perceptions of the tool before and after use and found that about 85% of students had a positive experience with the chatbot after the pilot phase.

This example is relevant because it demonstrates how a faculty first introduces AI on a limited scale (a pilot project in one discipline) to test user experience and pedagogical effects before broader implementation. Characteristics of good practice include: 1) user testing and evaluation (“before” and “after”), 2) adaptation of the tool (chatbot) to the specific needs of students in an online environment, and 3) collection of quantitative and qualitative data on the tool’s impact. The results show a significant number of users with positive perceptions, which can encourage further, broader implementation. Institutions considering introducing AI can adopt a pilot model with clearly defined evaluation phases, involving end users and collecting data on experience and perception.