Listen and learn: Research paper Research papers are a pairing of two 18 minute presentations followed by 18 minutes of Discussion led by a Discussant, with remaining time for Q & A.
This is presentation 2 of 2, scroll down to see more details.
Other presentations in this group:
The use of Assessments and how they are being applied, and which are the main goals for applying assessements.
When we look at everyone involved in education - students, courses, teachers, and educational institutes - we focus on four key questions/axes:
Who: We identify who is gathering the assessment data.
When: We determine when during the learning process this data is collected.
How: We explain the methods used to collect this information.
What: We specify exactly what we're measuring or assessing.
In terms of assessing students, we concentrate on three main areas: their academic skills, their soft skills, and their psychological state.
By skillfully using a Large Language Model, alongside data collection and specific prompts, we generate a report that assesses our subject of study. In this research, we rely on the ChatGPT model as our primary tool.
Methodology
Our database relies on accessible tools - Google Forms and Google Sheets - to ensure user-friendliness. In the initial phase, they serve as our data repository. The development was carried out using Python programming language and Flask Framework to create the Web App, acting as an interface for the prompts and as a final report generator. The Google Sheets API facilitates seamless data handling, while the ChatGPT API empowers us to utilize ChatGPT through chat completions.
Data Collection and Storage for Assessment
For data gathering, we've designed forms on Google Forms. These forms are straightforward and simple, allowing us to gather assessment data. The forms data can be provided by parents, students, educators, as well as external collaborators like psychologists and special education teachers. The results from each form are automatically stored in Google Sheets.
The Language Model
We've employed the ChatGPT model through its API, utilizing chat completions. We'll be starting with version 4, without any fine-tuning (there is no option at this time of writing), in the initial phase of pilot testing.
By aiming to help all subjects in the educational process progress and improve, from students to educators, and the content and methodology of teaching, we demonstrate how Artificial Intelligence can be employed to assist humans without sidelining them. This paves the way for similar applications. The strength of such a model lies in its potential to propel educational assessment into a new era, one we have not witnessed before. The goal is to enhance and advance all those involved in the educational process, with assessment being the most crucial aspect of education.
In this study, significant emphasis is placed on prompt engineering for each stage of questioning in chat completions. During the pilot phase, we will explore what prompts should be used to yield the best results. Additionally, the token limits in prompts and responses from ChatGPT might turn out to be an advantage, focusing more on the model's engineering, storing data of higher significance, and creating better prompts, rather than adopting a "throw everything in and prompt" approach.
During the pilot phase, we intend to expand the scope of assessed data to include information from wearable devices, offering insights into the physical presence of the student and teacher through biometric data, after the evaluation of which information should be obtained. Furthermore, we aim to integrate this data with information from other electronic classrooms/data for a more comprehensive assessment, especially in blended learning scenarios, effectively supporting both learning approaches in the best possible way.
Finally, the grander vision regarding the model's application (in this phase) involves connecting data and assessments from the same or various types of educational institutions for an even more insightful comparative assessment. This could have profound implications for understanding learning on a geographical basis.
While the data and information from the pilot model are restricted to the location they were collected and generated, special attention must be given to privacy and where the data and information related to the assessment go.
Artificial Intelligence (AI) stands as the cornerstone of future technologies, representing a pivotal force akin to a general-purpose technology. As researchers diligently strive to harness its potential across diverse applications and tasks, the profound impact on our daily lives is imminent, promising a sweeping transformation in various domains.
Education, in particular, is a focal point of research in terms of how Artificial Intelligence, - Artificial Intelligence in Education (AIED) - can revolutionize its multifaceted dimensions. The transformative potential of AI in education encompasses the operation of educational institutions, the pedagogical interaction between learners and courses, the content and methodology of instruction, among other facets.
Artificial Intelligence inTechnology-Enhanced Assessment: A Survey of Machine Learning
A system for formative assessment and monitoring of students’ progress.
Conducting psychological assessments in schools: Adapting for converging skills and expanding knowledge
TEACHERS’ GUIDE TO ASSESSMENT
Formative assessment: Teacher knowledge and skills to make it happen
The Teachers’ Assessment Knowledge and Practice: Contribution of the Past-Time Experiences to the Present-Time Decision
ChatGPT as an Educational Tool: Opportunities, Challenges, and Recommendations for Communication, Business Writing, and Composition Courses
Guest Editorial: Precision Education - A New Challenge for AI in Education
Application and theory gaps during the rise of Artificial Intelligence in Education
Return
Trips and Tours