better-ed AI

At better-ed, we believe the quality of an assessment depends on the quality of its questions. So we set out to answer a simple question: how do you measure whether a question is actually good for a given student?

Three goals drove this research:

Ask the right questions. We want better-ed to ask questions calibrated to each student's learning capability, not one-size-fits-all.
Go beyond memorization. Traditional exams mostly test recall. We want to challenge students to analyze, evaluate, and create. Bloom's Taxonomy gives us a proven framework for measuring that kind of higher-order thinking.
Make questions easy to understand. A great question is useless if a student can't parse it. We use CEFR levels to calibrate the language complexity of each question to the student's age and grade level.

The result: a model that scores questions on both cognitive depth (Bloom's level) and language complexity (CEFR level), so better-ed can ask questions that are challenging in the right way and clear enough to answer.

Link to paper in the comments.

We'd love to hear from educators and researchers working on assessment. What makes a question great in your classroom?

🎉 𝗯𝗲𝘁𝘁𝗲𝗿-𝗲𝗱 𝗶𝘀 𝘀𝗵𝗼𝗿𝘁𝗹𝗶𝘀𝘁𝗲𝗱 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗤𝗦 𝗥𝗲𝗶𝗺𝗮𝗴𝗶𝗻𝗲 𝗘𝗱𝘂𝗰𝗮𝘁𝗶𝗼𝗻 𝗔𝘄𝗮𝗿𝗱𝘀!

We're thrilled to be recognized by one of the most prestigious awards in global education, chosen from 𝟮,𝟬𝟬𝟬+ 𝗲𝗻𝘁𝗿𝗶𝗲𝘀 worldwide.

In the words of QS Reimagine Education:

"This shortlisted application demonstrated exceptional innovation, impact, and scalability, successfully progressing through four rigorous evaluation rounds conducted by a panel of 1,634 international higher education and edtech experts."

Since launch, better-ed has saved teachers 𝟮𝟴,𝟬𝟬𝟬+ 𝗺𝗶𝗻𝘂𝘁𝗲𝘀 while keeping 𝘁𝗲𝗮𝗰𝗵𝗲𝗿𝘀 𝗶𝗻 𝗰𝗼𝗻𝘁𝗿𝗼𝗹 and every assessment focused on 𝗴𝗲𝗻𝘂𝗶𝗻𝗲 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴, not just right answers.

Thank you DOST PCIEERD for the trust and support.

🌍 W𝗲'𝗿𝗲 𝗴𝗹𝗼𝗯𝗮𝗹, from the Philippines to Europe and beyond. On to the finals! 🙌

api.accredible.com

OECD's PISA 2025: AI can help learning when it's purposeful, moderate, and guided.

Students who used AI to draft writing often scored lower than those who didn't.
But frequent users who used AI to *learn*, and were taught how to check AI output, tended to do better than users with no guidance.

AI as a substitute for thinking hurts.
AI as a scaffold for thinking can help.

The teacher's job isn't catching AI. It's knowing whether students understood.

better-ed helps you hear the reasoning, in their own words.
Free for teachers → better-ed.ai

Source: OECD PISA 2025 Results (Volume I)
https://www.oecd.org/en/publications/2026/09/pisa-2025-results-volume-i_5265bfb1.html

Good news for Philippine classrooms.

PISA 2025: Filipino students posted their highest scores yet in reading, math, and science.

vs 2022: Reading +20 · Math +16 · Science +17

More learners hit baseline (Level 2+): reading 24%→31%, math 16%→21%, science 23%→32%. OECD flagged PH among the fastest improvers in reading.

Progress is real. Practice makes perfect.
In our math pilot, students who reviewed and practiced with better-ed improved by up to 21 points on average.

Free for teachers → better-ed.ai

Two students get the same answer right.

One understands it. One memorised it.

From the answer alone, you can't tell which is which — and that's the problem better-ed was built to solve. A short spoken explanation tells you in five minutes what a paper quiz can't.

See it free → better-ed.ai

"It helps us see beyond students' answers."

That's Dean Aris Ignacio of Southville International School and Colleges, on what changed for his faculty.

A score tells you if a student was right. It doesn't tell you how they got there — or where they got lost.

When teachers can see the thinking, they can go back to doing what they do best: guiding, encouraging, helping every student grow.

Salamat po, Dean Aris. 🇵🇭