A very informative publication from SURF (The Netherlands) is Promises of Al in Education. Discussing the impact of Al systems in educational practises from J. Walker and D. Baten, June 2022.
In this post, I present some key points from the report.
- The main goal of the publication is to give educational professionals a starting point from which they can get more acquainted with possible applications of Al in education.
- When Al systems are introduced into education, new and often challenging pedagogical options arise. An important question is whether the options offered sufficiently support education or, whether the quality of the educational process for students and teachers strengthens and whether the options are feasible for the teacher and students.
The field of education needs to develop an increasingly better understanding of Al in education. The authors state that the teachers, researchers and developers of Al systems must work together to develop and implement Al in education. Educational institutions still have a unique position where they can participate in discussions about the desired future effects and applications of technologies inside and outside the classroom. Without these interactions, education professionals will always fall behind.
The educational professionals should indicate more than now what you could and would like to use Al to improve education. In this way, they can positively influence the direction within the development process of Al for education.
- There are different definitions of Al. An example is: “Artificial intelligence system (Al system), which means software that is developed with one or more of the techniques and approaches listed in the separate post and can, for a given set of human-defined objectives, generate outputs such as content, predictions, recommendations, or decisions influencing the environments it interacts with”.
- In the report, four Al applications are presented and discussed in detail. It provides a good and helpful insight into what Al can do for higher education:
Teachers use Prerusall to support and motivate students in preparing for classroom lectures. Perusall shows that an Al system can inform teachers about students’ reading skills and involvement in assignments. Students can be helped through automated feedback. They should get used to the reactions/judgments that Al generates in response to the student’s work. The AI is prepared based on a large number of examples. Teachers must identify possible gaps in the AI and fill in these gaps in the AI to keep the AI up-to-date.
The Jill Watson Al Agent is a “Conversational Agent” who is integrated into the education program to answer frequently asked questions from students and communicate with students in different ways.
There is no Al who can answer every question. Training data to prepare a Jill Watson conversational agent may then be limited to the educational program it is intended to support. If the AI’s training data quality is good, the agent achieves positive results. To scale and promote a conversational agent at the organizational level, executive administrators need to consider how technologies like these can impact social engagement in courses and what kind of data management or regulation may be necessary to expand the reach of the conversational agent effectively.
PhotoMath is a mobile application that uses a smartphone’s camera to scan and recognize mathematical equations; the app then displays step-by-step explanations on the screen. PhotoMath, a smart device tool for solving math problems, impacts classroom exercises and homework. This has led some teachers to question whether they should ban the tool, limit its use or, conversely, name and use its correct value in current education.
GPT-3 is the largest publicly disclosed language model ever. Technologies such as GPT-3 are becoming increasingly competent in generating coherent and readable texts. This can affect both students’ skill development and the educational programmes’ quality. Educational institutions will be challenged by this option of automatically generated texts to make good use of them.
A broader application of these technologies, GPT-3 and PhotoMath, for example, can impact students’ learning and skills development that have not been considered before. It is up to the education field to decide whether these new pedagogical options will be used in education and how this can be done.