Blog

How can policymakers and NGOS use ChatGPT for FLN

This article is the first in a series of articles on how ChatGPT and AI models can be leveraged for FLN.

Governments can use GPT-3, or other artificial intelligence models, in a number of ways to improve learning outcomes for students:

Generating educational materials: Governments can use AI models to generate
educational materials for foundational literacy and numeracy education. These materials
can be customised to meet the needs of individual students and can provide practice
opportunities that help students develop their skills.
Improving teacher productivity: Governments can use AI models to automate the
generation of educational materials and provide practice opportunities for students. This
can free up teachers’ time and allow them to focus on other aspects of their job, such as
individualised attention to students and assessment of their progress.
Personalising learning: AI models can be used to generate personalised educational
materials that meet the specific needs and abilities of individual students. This can help
to keep students engaged and motivated and can make the learning experience more
effective.
Supporting remote learning: Governments can use AI models to support remote
learning, which is becoming increasingly important due to the COVID-19 pandemic. AI
models can provide educational materials and practice opportunities for students who
are unable to attend traditional classes in person.
Providing real-time feedback: Governments can use AI models to provide real-time
feedback to students, helping to improve their literacy and numeracy skills. This
feedback can help students understand their strengths and weaknesses and target their
efforts more effectively.
Informing policy and decision making: Governments can use the data generated by AI
models to inform policy and decision making. This data can help to understand the
effectiveness of different teaching methods and identify areas where additional
resources or support may be needed.
Strengthening partnerships: Non-profits can use AI models to strengthen partnerships
with governments and other organisations, allowing them to collaborate more effectively
on initiatives aimed at improving education outcomes.

Parallely, there could be some issues, both for governments and non-profits working to improve learning outcomes in FLN.

Quality control: The quality of educational materials generated by AI models may vary,
and there is a risk that inaccurate or inappropriate information may be presented to
students. This could negatively impact their learning and create confusion.
Bias: AI models can perpetuate and amplify existing biases in society. This could lead to
educational materials that are not inclusive or culturally sensitive, which could have a
negative impact on students’ learning experiences.
Dependence on technology: The use of AI models for foundational literacy and
numeracy education may make students overly dependent on technology and reduce
their motivation to engage in traditional learning activities.
Data privacy: The use of AI models for foundational literacy and numeracy education
may raise concerns about student data privacy and security. Governments and
educators must ensure that data is collected, stored, and used in a secure and
responsible manner.
Limited creativity and critical thinking: While AI models can generate educational
materials and provide practice opportunities, they may not be able to fully replace
human teachers and may not promote creative and critical thinking as effectively.
Cost: The development and deployment of AI models for foundational literacy and
numeracy education may be expensive, and governments must carefully consider the
cost-benefit of such initiatives.
Technical limitations: AI models such as GPT-3 are not perfect, and may not be able to
fully understand context and nuances in language, which could lead to
misunderstandings and mistakes in the educational materials they generate..

Our next articles in this series will look into how AI models can be leveraged to generate content for teaching FLN and also show samples of what kind of content can be generated.

This article was written with the help of ChatGPT.

Banner image courtesy: An AI generated image by DALL – E

This article was written under the aegis of the Centre for Education Research in India (CERI). CERI, an initiative powered by Madhi Foundation, is a digital repository and think-tank catering to policymakers, practitioners, and academics in the education sector and the larger community, to catalyse reform in the education ecosystem in India.

em {font-style: italic; font-size: 16px;}

Posted by Vishal V
Blog

How do student’s fare in Oral Reading Fluency? A juxtaposition of the Foundational Learning Study and Madhi’s own assessment

Based on multiple reports including Annual Status of Education Report (ASER, 2014, 2018), National Achievement Survey (NAS, 2021) etc., it is now widely accepted that children in India, and in Tamil Nadu, are in the midst of a learning crisis which show that a large percentage of children in India are lagging behind in Foundational Literacy and Numeracy skills. In a crucial step towards strengthening efforts to bridge this gap, the Ministry of Education, Government of India has conducted a large-scale nationwide Foundational Learning Study (FLS) in collaboration with the National Council of Education Research and Training (NCERT, 2022). The study aims to set up benchmarks for reading with comprehension in 20 Indian languages and is one of the largest one-on-one studies with a sample size of 85000+ Grade 3 students. Parallelly, Madhi Foundation conducted a study that focuses on identifying the learning levels of children in Class 1–3 in Tamil, English, and Mathematics by taking a representative sample of 3600 students from the districts of Chennai, Ariyalur, Salem, and Thoothukudi.

In this report, we will analyse Oral Reading Fluency with reading comprehension from the National Report on Benchmarking for ORF and Numeracy and compare it with Madhi’s study to understand what level students in primary classes in Tamil Nadu are currently at.

According to the FLS, it was found that around 42% of students in Tamil Nadu could only read 0–8 words correctly in Tamil in a given period of time. Only 23% met or exceeded the global minimum proficiency standard of reading at least 28 words per minute when in grade 3. The findings from Madhi’s study corroborate the FLS on foundational literacy in Tamil. We found that the average number of words/phrases that a child in Class 3 could read was 9 words per minute with a maximum of 15 words across the sample data set. This is indeed a grave situation. However, the FLS study on English presents a slightly different picture. It studies almost all the states in India where there is English medium education and provides a national picture of foundational literacy, as compared to disaggregated state-based data. The FLS found that around 55% of students meet global proficiency standards of 35–53 correctly read words per minute. However, it is important to note that this is an India-wide average which could have severe variations across states. According to the study conducted by Madhi, a class 3 student read anywhere between 3 to a maximum of 15 words per minute in English.

In numeracy, the numbers are equally stark. According to the National Report, around 29% of all students in class 3 did not even partially meet global minimum proficiency standards. Only 20% of students even met the global minimum proficiency standard in numeracy.

Despite the rather bleak picture, it is indeed a positive sign that such benchmark studies are being conducted, which gives policymakers better data and insight from which they can design and implement interventions. Such benchmarking studies provide stakeholders with essential data to identify critical areas for improvement and develop appropriate action plans, which are often contextual as well. For example, in the Madhi conducted study, we found that the medium of instruction had an impact on the performance of students in many of the tested skills.

While this study focuses on benchmarking results and standards, it would be very useful to set up process benchmarks as well. These studies can be used to assess performance objectively while also providing contextual insights; expose areas where improvement is needed in Foundational Literacy and Numeracy; identify other states/countries with processes resulting in superior performance, with a view to their adoption; and most importantly, test whether improvement programmes have been successful and cost-effective.

Reference

Annual Status of Education Report (2014),

014mainreport_1.pdf

Annual Status of Education Report (2018),
http://img.asercentre.org/docs/ASER%202018/Release%20Material/aserreport2018.pdf

National Council of Educational Research and Training (NCERT) (2022), Foundational Learning Study, National report on benchmarking for oral reading fluency with reading comprehension and numeracy,
https://ncert.nic.in/pdf/FLS/fls_orf/ORF.pdf

Ministry of Education, National Achievement Survey (2021),
https://nas.gov.in/report-card/2021

This article was written under the aegis of the Centre for Education Research in India (CERI). CERI, an initiative powered by Madhi Foundation, is a digital repository and think-tank catering to policymakers, practitioners, and academics in the education sector and the larger community, to catalyse reform in the education ecosystem in India.

em {font-style: italic; font-size: 16px;}

Posted by Vishal V