Supporting Education in Cambodia with Socratic AI Learning

Project summary

Please contact me at steven.watson@projectascend.org with any questions.

In March 2026, we ran a 2 month pilot to study how AI tutoring shifts learning culture and class roles. We observed teachers taking full ownership, developing their own routines with confidence to train up other teachers. We observed students engaging deeper in their learning and becoming more active in the classroom, asking more questions not only in their AI classes but also in their classes outside of the pilot. We want to expand, running a pilot that observes the shifts over 6 months across a diverse range of schools.

The AI platform we introduce into the classroom, provides a low stakes environment for students to ask questions and learn through the Socratic method, this is particularly beneficial in regions like Cambodia where teacher subject knowledge is limited. The AI adjusts for individual students bringing student interests into subject explanations, helping students the struggle to concentrate engage with their learning.

We keep teachers as a central piece in the pedagogy, AI doesn't replace the teacher, but becomes a tool for them to adopt. The AI delivers meaning to the students, freeing the teachers up to mentor students, manage the class, and support students in how they learn. We aim to leave classrooms fully independent of us to ensure schools are self sustainable with this new method of learning.

We aim to run the expanded pilot from May to October 2027. However a lot of prep before and write up after will also be a part of this project.

What are this project's goals? How will you achieve them?

We want to run a second pilot across 4 schools with around 200 students. This pilot will explore the following:
Can AI learning be adopted by teachers in rural, public and private school environments in Cambodia?
How do test scores and academic ability shift over 6 months of AI learning?
How can student avoidance be managed when AI is introduces as a learning method?

The answer to these questions will be published in a paper. Beyond that we aim to refine the SEL framework for AI interaction grading (shown on our website), create an open source learning platform with integration documentation, and, assuming the pilot is tracking successfully, establish long term integration for schools to support learning with AI following the pilots conclusion.

This will be achieved by setting up in Cambodia for 6 months. Setting up infrastructure for schools that need it, training the teachers and supporting classroom activities while taking classroom observations, talking to teachers, tracking AI use and class tests to measure results. All work done in these schools will be done along Ethical guidelines with anonymized data to ensure student protection and wellbeing. The full methodology can sent upon request by emailing steven.watson@projectascend.org. Alternatively, you can view our past paper on our website to see our similar methodology for our last pilot.

How will this funding be used?

The funding will be split across the following:
Devices: $250 per laptop for around 25 to 80 laptops: $6250 - $20000
Infrastructure (internet/charging/storage for laptops): $1000 - $6000 (1-2k per school, 1-3 schools that need it)
Myself and translator living allowance for 6 months (food, accommodation, travel): $17,500
Teacher Compensation: $2000 - $4000
Independent Ethics Review: $500 - $2000
API tokens: $3000 (a lot of this cost is for message classification and analysis, post study costs will be much lower)
Contingency (laptops being stolen/broken): $1000 - $5000
Licenses & subscriptions: $500 - $2000
Headsets: $1600
Translator (Unpaid internship is last resort): $0 - $20000
Donor/Consultant visits: $0 - $3000
Personal Allowance (This will allow me to work full time on the project in the lead up and to write the paper): $0 - $20000

Total Cost Estimate:
$33350 - $104100

Trade offs for lower threshold:
Only 1 school without complete infrastructure can be included, the paper will be less credible when judging AI learning in developing classrooms.
Less compensation for teachers and lower quality ethics review
It will be difficult finding a translator to support as it must be volunteer without pay.
I will need to work alongside this project in the lead up and post, this risks the learning platform and integration pedagogy development, and prolongs the paper write up after.

Who is on your team? What's your track record on similar projects?

Steven Watson - Project Lead, data engineer and ran the previous pilot in Cambodia successfully
Dr Steve Watson (no relation) - Project Advisor, professor at University of Cambridge in the faculty of Education. Also advised on previous pilot.
Translator - tbd

Previously we ran a 2 month pilot with similar methodology successfully. More can be found on our website.

What are the most likely causes and outcomes if this project fails?

The AI classes are run alongside students regular class, so there is no risk of their academic ability being impacted. Worst case scenario, a few developing schools receive infrastructure to support their learning and we publish a paper detailing what went wrong to support future research in this area. Based on our previous paper it is unlikely for our pedagogy to fail.

How much money have you raised in the last 12 months, and from where?

Currently I have self funded this entire Project. $15k from my own savings went towards running the first pilot in Cambodia and sustaining myself to plan prior and publish the paper after. I am reaching out on Manifund as I have run out of savings and this pedagogy has proven to provide value, so I want to continue my work to improve education.