AI Just Beat Law Professors at Their Own Game

🎓 AI in Education Digest

AI outscored law professors in a blind Stanford test, a Gemini tutor moved real maths scores in Sierra Leone, and UK universities want AI literacy baked into every course. Here's your 5-minute roundup of what actually changed in AI and learning today.

By Dereck Tafuma · June 20, 2026 · 5 min read

Story 01

Law professors preferred AI answers in 75% of blind tests

Source: EdTech Innovation Hub (Stanford-led study) · Published June 19, 2026

Here's a number that should stop every "AI can't really teach" argument in its tracks. Sixteen professors across 14 US law schools ran 2,918 blind comparisons between answers written by fellow instructors and answers from Google Gemini 2.5 Pro and NotebookLM. They were asked a simple thing: which answer would you rather give a student? The machines won 75% of the time.

It gets better. The AI answers were flagged as potentially harmful to learning in just 3.5% of cases — versus 12% for the human-written ones. And these weren't trivia questions; many required synthesising messy legal doctrine and weighing competing arguments with no single right answer.

Why it matters: This is the first serious study to test AI on judgment-heavy questions where there's no clean answer key — exactly the kind of explanation a student needs at the kitchen table. It reframes the debate from "is AI accurate enough?" to "how do we use it responsibly?"

My take: the professors didn't lose to a calculator — they lost to a patient explainer that never has a bad day. That's the part students should pay attention to.

Story 02

A Gemini tutor lifted real maths scores in Sierra Leone

Source: EdTech Innovation Hub (Google DeepMind & Fab AI RCT) · Published June 16, 2026

Forget the hype — this is a proper randomised controlled trial. 1,763 Grade 7 and 8 students across 12 schools in Sierra Leone used Gemini Guided Learning during teacher-led maths lessons. The result: a 0.258 standard-deviation jump in scores, roughly the equivalent of moving from the 50th to the 60th percentile, or more than a year of extra learning progress.

The crucial detail is how it was run. Teachers stayed in charge, students worked in pairs (a "driver" typing and a "navigator" steering the questions), and the AI was told to coach, not hand over answers — 76% of its messages were scaffolding questions, only 2% were solutions. Students who hit the full 12 hours saw the biggest gains.

Why it matters: It's hard evidence that AI tutoring works when it's wrapped around a teacher, not used to replace one. But the researchers flagged a warning: stronger students gained the most, so AI could widen attainment gaps if we're not careful.

My take: "AI plus a good teacher" is the winning formula here — the tech was a tool in the lesson, not the lesson itself.

Story 03

Leicester pushes for AI literacy in every core curriculum

Source: EdTech Innovation Hub (University of Leicester) · Published June 17, 2026

The University of Leicester is calling for AI literacy to be built into core curricula rather than bolted on as an optional extra, pointing to its own research linking structured reflection on AI use to stronger critical thinking. The argument is simple: if students are going to use these tools anyway, the skill to teach is judgment — knowing when to trust, question, and double-check the output.

Why it matters: Banning AI is losing ground to teaching it well. Embedding literacy across subjects is how universities turn "students are cheating" panic into "students are thinking" — and it sets the expectation employers now have.

My take: the schools that win the next decade won't be the ones that block AI — they'll be the ones that teach you to interrogate it.

Story 04

Durham wins £25k to expand AI assessment work in Irish schools

Source: EdTech Innovation Hub (Durham University) · Published June 18, 2026

Durham University researchers have secured £25,000 in funding to extend their work on AI in assessment across Irish schools. It's a smaller headline than a Stanford study, but it points at the question every teacher is quietly wrestling with: how do you assess learning fairly in a world where a chatbot can draft an essay in seconds?

Why it matters: Assessment is the pressure point for AI in schools. Getting it right — through better task design, not just detection software — is what keeps qualifications meaningful.

My take: the future of marking isn't "catch the cheats," it's "design work that AI can't do for you." This funding is a small bet on exactly that.

Story 05

68% of young students say AI saves time — but 1 in 5 feel less creative

Source: EdTech Innovation Hub (student survey) · Published June 12, 2026

A new survey of young students found 68% say AI saves them time — yet roughly one in five report feeling less creative when they lean on it. That tension is the whole story of AI in education right now: a genuine productivity boost shadowed by a quiet worry that the thinking is being outsourced.

Why it matters: Speed isn't the same as learning. If students offload the hard, creative parts to AI, they bank time now and lose skill later — which is exactly the trap parents and teachers need to watch for.

My take: use AI to remove the friction, not the thinking. The students who thrive will treat it like a sparring partner, not a ghostwriter.

Story 06

MDPI turns on AI integrity screening across 2,000 daily submissions

Source: EdTech Innovation Hub (MDPI) · Published June 18, 2026

Academic publisher MDPI has begun running AI-powered integrity screening across roughly 2,000 manuscript submissions a day, using automated checks to flag potential research-integrity issues before papers reach review. It's AI being used to police the very flood of AI-assisted writing now hitting journals.

Why it matters: The same technology students use to write is now being used to vet what gets published. For anyone heading into higher ed or research, AI literacy and academic integrity are becoming two sides of the same coin.

My take: AI is now on both sides of the desk — writing the work and checking it. Knowing the rules of the game has never mattered more.

🎬 Best story for today's video