Hi PH 👋 I’m from Uzbekistan. An IELTS attempt here costs close to a month’s average salary, so people can’t afford to find out on test day that they were a 6.0 and not a 7.0. But self-preparation breaks down exactly where it matters: Reading and Listening you can check against an answer key, and Writing and Speaking you can’t check at all. I started with a Telegram bot that graded essays – 1,200+ of them. The feedback requests taught me people don’t want a number, they want to know which sentence cost them a band. So I built the full thing. BandCheck covers all four sections. Reading and Listening with timed practice and explanations for wrong answers, Writing with per-criterion bands and line-level rewrites, Speaking with recorded responses assessed on fluency, pronunciation, and lexical range. What I’m still working out - how much examiner judgment is genuinely reproducible. Rubric compliance, yes. “Sophistication of expression” I’m not convinced yet. If you’ve taken IELTS or teach it, I’d really like your read on where the model is too generous.

Forum Threads

p/bandcheck

•

2d ago

How much of IELTS examiner judgment do you think is actually reproducible by a model

I ve spent two years building AI scoring for IELTS, and I still don t have a clean answer to this.

Here s where I ve landed so far. Some parts of the rubric are mechanical and a model handles them well: word count, task coverage, paragraph structure, grammatical range, whether the response actually answers the prompt. Reading and Listening are trivially checkable there s a key.

View all

BandCheck

Practice all 4 IELTS sections with an AI examiner

Practice all 4 IELTS sections with an AI examiner

Forum Threads

How much of IELTS examiner judgment do you think is actually reproducible by a model

Forum Threads

How much of IELTS examiner judgment do you think is actually reproducible by a model